A Gathering of Eagles: Design evolution of the F-22 Raptor


Lockheed had done the first studies by the early 1970s for a ‘superstealth’ air-to-ground attack airplane for the U.S. Navy. Lockheed used superstealth to refer to a significant improvement in an aircraft’s stealth characteristics.
Bart Osborne, program manager for Lockheed Tactical Systems in 1972

A significant portion of the history of the F-22 spent years encased in a collection of wooden boxes stacked in a small storage closet in the engineering building at Lockheed Martin Aeronautics Company in Fort Worth, Texas.

The containers conceal a variety of design study models dating back to the earliest phases of the Advanced Tactical Fighter-to what eventually became the F-22 Raptor.
The official beginning of the ATF program usually traces to 1981 when the U.S. Air Force’s Aeronautical Systems Division, or ASD, released a request for information for concepts for an advanced tactical fighter. ASD is now the Aeronautical Systems Center at Wright-Patterson AFB, Ohio.

The term advanced tactical fighter and its abbreviation, ATF, however, appeared in a general operational requirements document issued to contractors by the Advanced Planning Branch of ASD almost ten years earlier in 1972. The requirements document pertained to a new air-to-ground fighter to complement the new F-15 air-superiority fighter. The ATF would replace an aging fleet of F-4 and F-111 aircraft.

ASD awarded concept exploration contracts for this ATF to General Dynamics and McDonnell Douglas. The requirements called for an air-to-ground fighter that could fly Mach 2.5 at high to medium altitudes and carry standoff weapons designed to destroy tanks and other ground targets. That aircraft never materialized. The F-16 Fighting Falcon, originally designed as an air-to-air day fighter, came in the back door to fill the air-to-ground role. ASD would have to wait ten years to embark on another new fighter program.

Studies for a new fighter subsequently shifted away from ASD to the Air Force Flight Dynamics Lab, also located at Wright-Patterson (and now called the Air Vehicles Directorate of the Air Force Research Laboratory). While ASD supported new aircraft development programs, the lab pursued technologies related to military aircraft.
The Flight Dynamics Lab nurtured the advanced tactical fighter through the 1970s, sponsoring research and development contracts.

General Dynamics and McDonnell Douglas performed a 1975 study titled, “Advanced Technology Ground Attack Fighter.” After that, six companies participated in an “Air-to-Surface Technology Study.” The lab sponsored two more studies in 1980: a “Tactical Fighter Technology Alternatives” for future air-to-ground fighters and a “1995 Fighter Technology Study” for future air-to-air fighters. Boeing and Grumman conducted the air-to-ground studies. General Dynamics and McDonnell Douglas conducted the air-to-air studies.

Lockheed was also involved in early studies, but for other government agencies.

“Lockheed had done the first studies by the early 1970s for a ‘superstealth’ air-to-ground attack airplane for the U.S. Navy,” explains Bart Osborne, program manager for Lockheed’s Tactical Systems in 1972. Osborne, now retired, was chief engineer during the demonstration/validation phase of the ATF program in the mid-1980s. Lockheed used superstealth to refer to a significant improvement in an aircraft’s stealth characteristics.

“Lockheed’s early work led to the company’s proposal for the Navy’s Advanced Tactical Aircraft program,” Osborne continues. “It also led us to an air-to-ground USAF version, which was a superstealth design. We didn’t know how we were going to do superstealth in those early days. We had some ideas, but nothing proven. The operational research in these early studies showed us just how powerful superstealth could be.” Early ATF work at Lockheed, however, became dormant as the company focused its talents and energy on the early stages of the F-117 program.

Boeing studied a wide range of advanced air-to-ground fighters in the 1970s. “The aircraft ranged from relatively small single-engine airplanes to large twin engine airplanes,” recalls Dick Hardy, who was in charge of a preliminary design group for Boeing at the time. Hardy would later become program manager for the F-22 at Boeing. “Our designs also ranged from supersonic to subsonic,” he continues. “They varied from conventional non-low-observable airplanes to really low-observable airplanes and flying wings similar to the B-2 bomber.”

General Dynamics studied a wide range of advanced fighter concepts and modifications to existing fighters. Advanced derivatives of the F-16, F-15, and F-111 competed with the new concepts for the same missions. The advanced concepts included a conventional aircraft called “Plain Jane,” a supersonic stealth configuration, a small inexpensive fighter called “Bushwhacker,” a large fighter called “Missileer” that could carry many long-range air-to-air missiles, and a highly stealthy all-wing fighter called “Sneaky Pete,” which eventually evolved into the Navy’s short-lived A-12 Avenger II.

These and other government-funded studies were interspersed with company-funded studies at Boeing, General Dynamics, Lockheed, and at all the other companies that would eventually vie for the formal ATF program.

The underlying motivation for all of these studies was to identify the most promising design concepts and enabling technologies for potential missions and roles of the next generation fighter. The development of a new generation of fighters in the Soviet Union intensified these efforts in the United States. In the early 1970s, the Soviets were working on what became the MiG-29 and the Su-27. The prototype for the MiG-29 first flew in October 1977. The Su-27 prototype (the second prototype, which more closely resembled the final design) first flew in April 1981. The Soviets were also rapidly advancing their surface-to-air and air-to-air missile technology. Projected exchange rates between US- and Soviet-built fighters were looking unacceptably even to USAF planners.

In performing these studies in the 1970s, the aerospace industry developed a broad range of aircraft concepts. Companies also created complex computer models for rapidly evaluating specific designs based on these concepts. General Dynamics, for example, had a highly refined process for evaluating designs. The process began with a design concept, which defined a general arrangement and a suite of aerodynamic, structural, avionics, armament, and propulsion technologies. Synthesis and sizing computer models then produced families of designs having a broad range of maneuver, speed, range and other capabilities.

Families of these more specific designs were in turn fed into life-cycle cost models and into a set of effectiveness models that determined the susceptibility of each design to surface-to-air and air-to-air threats. Other models ascertained each fighter’s lethality against its intended targets. The data from the effectiveness models were used in campaign models that accounted for force structures, mission allocations, basing concepts, threat distributions, strategies, and other details that define theater-level scenarios. The campaign models fought wars in which each conceptual design was only one element of the total air forces. Each design was placed in the campaign in numbers proportional to its cost.

This process allowed engineers to see how performance levels or design features affected the military usefulness of a new aircraft. Technologies and design features could then be recommended according to their contribution to the overall effectiveness of the combined air forces in the theater. The results also highlighted the performance requirements that each design concept needed to maximize its cost-effectiveness.

Many of the analyses indicated stealth to be a highly desirable design feature. The final General Dynamics report in the “1995 Fighter Technology Study” (produced in 1980) identified stealth as the preeminent characteristic in achieving air superiority. “Fighters have almost always been camouflaged, and fighter pilots have always employed stealthy tactics,” explains Bill Moran, then General Dynamics program manager for many of the analyses for the Flight Dynamics Lab. Moran is now a deputy director on the F-22 program in Fort Worth.

“The ‘Red Baron Study,’ an Air Force real-time analysis of combat in Vietnam, showed that over half of the aircrews shot down and about eighty percent of those fired on in Vietnam were unaware of their attackers,” Moran continues. “Available data shows that same experience to be true in World Wars I and II and in Korea. Biographies of fighter aces throughout history are full of references that credit their success to an ability to see their opponents before they were seen themselves, attacking from out of the sun, and attacking from their adversary’s blind spot.”

Stealth, in this sense, means being able to detect the other guy before he detects you. In a narrower sense, the term applies to the application of various materials and techniques that significantly lower the susceptibility of an aircraft to being detected.

As detection techniques evolved, so have techniques for avoiding detection. The development of effective radar networks in the early days of World War II, for example, led to stealth counters to radar. Radar-absorbing materials were initially used by Germany on submarine periscopes and air-breathing U-boat snorkels in the 1940s. Germany later incorporated radar-evading shapes and materials into designs for a jet-propelled flying wing fighter-bomber derived from the Horten Ho IX. Radar-evading techniques were simultaneously developed in the United States.

Soon, stealth was applied to drones and missiles. The Ryan Q-2 Firebee and Lockheed D-21 drones incorporated radar-absorbing materials. Boeing applied stealth technologies in the late 1960s to its design for the Short-Range Attack Missile, more commonly known as the SRAM and designated the AGM-69. The supersonic SRAM was used on B-52 and B-1 bombers until 1990. “The missile met a stringent radar cross section requirement,” says Hardy, who worked on the program for Boeing.

General Dynamics was also a major player in the early days of stealth. The company built the prototype for the Air Force’s first radar antenna target scattering facility (known as RatScat) in White Sands, N.M. The facility was used to measure accurately the radar cross section of aircraft. General Dynamics built RatScat and operated it for the Air Force through the 1970s. The company also operated its own radar measurement facility in Meridian, Texas, through the 1980s.

In the late 1950s, General Dynamics pursued a highly stealthy design concept to meet requirements set by the Central Intelligence Agency for a supersonic high-altitude reconnaissance aircraft to replace the Lockheed U-2. The design began as a B-58 parasite known as Super Hustler. It evolved into an independent aircraft optimized to cruise at 125,000 feet at a speed of Mach 6.25. This configuration, dubbed Kingfish, was to be built mostly of pyro-ceram (a heat-resistant and radar-attenuating ceramic material). Two Marquardt ramjets powered the aircraft in the cruise portion of its mission. Two retractable General Electric J85 turbojets provided power for takeoff and for acceleration to speeds at which the ramjets could be ignited.

The radical General Dynamics design, however, lost out to its competitor from Lockheed in August 1959. The resulting Lockheed airplane, the single-seat A-12, was the forerunner of the two-seat USAF aircraft more widely recognized as the SR-71 Blackbird. The A-12 capitalized on radar-evasive shaping and radar-absorbing structural materials. It is credited as the first operational aircraft to incorporate stealth to a high degree in its original design. Canted tails, sawtoothed structure, pie-shaped panels on leading and trailing edges, blended wings and chines, and radar-absorbing structure and paint combined to reduce the aircraft’s radar cross section to a small fraction of more typical aircraft of the period.

Lockheed took stealth technology to an even higher level in the 1970s when it combined computer technology with some obscure mathematical formulas relating to the reflection of electromagnetic radiation. The resulting computer program, called Echo, could accurately predict the way a flat-surfaced object would appear on radar. The company applied the software in a Defense Advanced Research Projects Agency study to create the Have Blue aircraft. The small faceted aircraft, with tails canted inboard, was the predecessor of the F-117 Nighthawk. The development contract for the F-117 soon followed the DARPA project. The success of the F-117, significantly stealthier than any aircraft that preceded it, would greatly influence the ATF program.

The conceptual work generated for the Flight Dynamics Lab identified speed as another critical characteristic of an air-superiority fighter. Speed steals reaction time from the enemy and provides more freedom to engage and disengage as desired. The initiative often goes to the faster combatant.

Speed in these studies took the form of supercruise-supersonic flight without using an afterburner (a source of undesirable and unstealthy infrared energy). Optimizing an aircraft for supercruise leads to long, slender configurations with small highly swept wings and large high-temperature engines.

The ATF would not be the first military aircraft capable of supercruising. This title belongs to the B-58 Hustler. The B-58, however, had to employ its afterburners or dive steeply to accelerate through the transonic drag to get to the flight condition where it could supercruise. The F-16XL and newer-model F-16s are capable of supersonic flight without afterburner as well.

Maneuverability was identified as another important characteristic of air-superiority fighters. Unlike stealth and supercruise, high maneuverability is more often used as a defensive tactic rather than an offensive one. “Despite the popular image conveyed in books and movies, like Top Gun,” explains Moran, “outmaneuvering the other guy to shoot him down is usually not a good idea. Maneuvering engagements take too long and make you predictable to an opponent other than the one you’re after.

“Down through history, successful air-to-air pilots have generally avoided maneuvering engagements whenever possible,” Moran continues. “Sometimes they forgot, however. Manfred von Richthofen, the famous Red Baron, achieved a record-high eighty-three kills in World War I, only to be shot down himself while violating this precept. Fighter pilots usually choose to operate in parts of the flight envelope that favor maneuverability characteristics of their own aircraft and avoid areas where their opponents have an advantage. Having good maneuverability everywhere, however, eliminates that consideration and thus increases tactical flexibility.”

Maneuverability was quantified in the 1960s by John Boyd in his energy-maneuverability theory [see “Tribute to John Boyd” in the July 1997 issue of Code One]. Boyd’s ideas were used on the F-15, but the F-16 was the first airplane to be designed specifically to emphasize the principles established by this theory. The most common measures of merit for energy maneuverability are sustained g capability (the ability to turn hard without losing airspeed and altitude); instantaneous g (the ability to turn the nose of the aircraft without regard to the effect on speed); and specific excess power (a measure of an aircraft’s potential to climb, accelerate, or turn at any flight condition). Another parameter of interest is transonic acceleration time (for example, the time needed to go from Mach 0.8 to Mach 1.2). Comparing these characteristics for two fighters shows which one should have the tactical advantage in a maneuvering engagement.

The ability to operate an aircraft from battle-damaged runways was yet another characteristic evaluated in the early ATF studies of the 1970s and 1980s. Designs incorporating this capability are referred to by a number of terms, including short takeoff and landing, short takeoff and vertical landing, and vertical takeoff and landing (STOL, STOVL, and VTOL, respectively). The benefits of abbreviated takeoffs and landings are, however, less clear than benefits associated with stealth, speed, and maneuverability.

Incorporating a short or vertical takeoff and landing capability comes at a high price. Establishing the need for the capability is very complex. “How much of this capability is needed depends on the size and effectiveness of an enemy’s offensive counter air [the ability to destroy adversary aircraft before they take off] and runway-busting weapons,” Moran explains. “It also depends on the number of runways that an enemy needs to wreck to significantly affect the other side’s ability to fly and on its ability to quickly repair runways and get back into operation. Most importantly, it depends on the effectiveness of defensive counter air forces in stopping enemy sorties from reaching a base in the first place.”

Whether to bomb concrete or whether to bomb airplanes on the ground further complicates the issue. The effectiveness of aircraft shelters and the accuracy of intelligence about where airplanes are at any given time play a role in the latter approach. “The United States and its NATO allies were busy developing a variety of air-to-ground weapons for busting large expanses of concrete,” Moran continues. “We reasonably assumed that the Warsaw Pact would reply in kind.”

“Short Snort” and “Jiminy Cricket” were two General Dynamics designs that addressed short takeoff and landings more directly. Short Snort vectored thrust from the engine over the wing to produce a fighter with runway requirements of only a few hundred feet. The concept employed ducts and ports that diverted engine exhaust out spanwise along the top of the wing. The approach generated tremendous amounts of lift at very low speeds. The ducting, however, was very heavy, and ultimately proved impossible to incorporate in a high-performance supersonic fighter.

Jiminy Cricket attacked the short takeoff and landing problem with multiple engines. The design had a main lift-cruise engine that provided thrust for lift and for up-and-away flight, and it had auxiliary engines mounted vertically that provided lift for takeoffs and landings only. The aircraft could have either short or vertical takeoff capability, depending on the size of the engines.

A fighter designed for supersonic flight and high maneuverability has a thrust-to-weight ratio and wing loading that produce a fair amount of inherent short airfield capability. That capability can be improved with the addition of rough-field landing gear (for operating on repaired runways), oversized brakes, and thrust reversers. The weight of the improvements, however, decreases the thrust-to-weight ratio available in air combat.

ATF originally had a very difficult STOL requirement that called for the use of some of these features, most notably thrust reversing and thrust vectoring. The takeoff and landing distances were relaxed during the demonstration/validation phase of the program to eliminate the need for reversers as well as their extra weight and cost. Thrust vectoring was retained since it improves aircraft performance in several ways. Thrust vectoring can be used to shorten takeoffs by rotating the nose of the aircraft up at a lower speed than would be possible by using tail surfaces alone. In cruise flight, vectoring can be used to supplement the trim normally provided by tail surfaces. Vectoring, therefore, allows smaller tails or allows cruise with tails set to a position that produces less drag. Thrust vectoring can also augment control power at high angles of attack or during aggressive maneuvers.

Lockheed’s early configurations featured faceted designs based on the F-117.

The challenge
Integrating stealth, speed, and maneuverability became the fundamental challenge of the ATF program. No one had ever attempted such a complex combination before. As the F-117 had shown, stealth affects every aspect of a design. Internal weapon carriage, a must for a stealthy design, increases the cross section of an airplane. Larger cross sections increase supersonic drag and work against supercruise. “A stealthy airplane requires a big weapon bay,” explains Hardy. “And the landing gear and the inlet duct want to be in the same place as the weapon bay. You wind up with a guppy that won’t go supersonic unless you make it very long with huge engines. Such an approach is a nonstarter because the airplane would be way too expensive.”

Maneuverability requirements tend to increase the size of the wings and tails and make the engines bigger than necessary for supercruise alone, all of which make stealth more difficult to achieve. Those few pilots who were briefed on the F-117 knew about compromises in speed, maneuverability, payload, and other capabilities that went along with an all-out approach to stealth. Fighter pilots who would be flying the ATF would not willingly sacrifice these capabilities for stealth.

Nine airframe companies and three engine manufacturers responded to the challenge when ASD reentered the game and issued its request for information for an ATF in 1981. At this early stage in the program, the Air Force had not decided whether the new aircraft would emphasize air-to-air or air-to-ground missions. The Air Force invited industry to share ideas for a new fighter.

The companies submitted a wide range of configurations in their responses. Lockheed’s response favored a derivative of the YF-12A (what most people would recognize as a single-seat SR-71). This aircraft, designed for air-to-ground missions, carried several kinetic-energy penetrator weapons in a central weapon bay. The weapons would be released at supersonic speeds at high altitudes and guided by a laser. The approach, which was worked through spring 1982, built upon technical data gathered from a series of air-to-air missile launches from the YF-12A conducted in the mid- to late-1960s. The YF-12A had fired seven Hughes AIM-47 missiles at altitudes up to 80,000 feet at speeds over Mach 3. The shots, at aerial targets at ranges of over thirty miles, were highly successful. The high-altitude, high-speed approach was also one of Lockheed’s candidates for the F-X program, what became the F-15.

Like Lockheed, Boeing took a supersonic approach in replying to the request for information. The designs favored air-to-ground missions. “After studying a broad spectrum of airplanes, including flying wings, canards, four-tailed airplanes, two tails, side inlets, and nose inlets, Boeing homed in on a design fairly quickly,” recalls Hardy. “We thought the aircraft should be designed for higher speed, so we concentrated on designs with a higher fineness ratio.

“It was also obvious that we needed a good maneuvering airplane,” Hardy continues. “When the prime mission of the airplane later shifted to air-to-air, we quickly got rid of those things that did not have good control authority.” Boeing also stressed stealth with clever internal arrangements and weapon bay designs that carried munitions semi-submerged.

The response from General Dynamics favored two of the four concepts originally developed in the 1976-78 studies for the Flight Dynamics Lab. One was a Plain Jane derivative called Model 21. This design was a forerunner of the conventional family of configurations the company would explore in the next phase of the ATF program. Model 21 looked like a traditional member of the modern fighter family, but it wasn’t totally conventional. It had frontal shaping and treatment to reduce its radar cross section, strut-braced wings, a rotating nose that combined a radar with an infrared search and track system. Composite materials comprised forty percent of the aircraft structure. Its air-to-ground loadings included glide bombs with square cross sections.

The other candidate was a descendent of Sneaky Pete. General Dynamics, however, was not allowed to show USAF officials actual drawings of this design because of its classification. The company substituted surrogate drawings of a notional fighter that USAF officials soon dubbed “the marshmallow.” The real design was the starting point for all-wing studies explored in the next phase of the program.

After a year of study and report writing by industry, ASD performed mission analyses on four generic fighter designs that spanned the variety of aircraft investigated by the companies. The aircraft were labeled N, SDM, SLO, and HI. N (numbers) was a small, cheap concept that could be bought in quantity. SDM (supersonic dash and maneuver) emphasized speed and maneuverability. SLO (subsonic low observables) was based on a flying wing design. HI (high-Mach/high-altitude) represented a large missileer. The results, which were presented to all participants, favored the flying wing. The more conventional SDM fighter placed second in effectiveness. The missileer and inexpensive minifighter did not rate well in the analyses.

Momentum and funding
As the RFI results were announced, the ATF program gained momentum and funding. A mission element need statement, a required document that characterizes a particular mission, was issued in late 1981. Tactical Air Command (which later became part of Air Combat Command) created a corresponding statement of need-another required document that addresses threats, theaters of operation, and capabilities needed to accomplish the mission described in the mission element need statement. The statement of need is largely credited to Col. Mike Loh, the deputy commander of the requirements branch at TAC Headquarters (he later became a full general and ACC commander). The fifty-page document was provided to industry for comment in the summer of 1982. The requirements formally made the ATF a replacement for the F-15 in the air-superiority role. An ATF System Program Office was formed at Wright-Patterson AFB in 1983, and Col. Albert Piccirillo became its first director.

A request for proposals for the ATF engine was issued in May 1983. General Electric and Pratt & Whitney were awarded contracts to build and test competing engine designs. The General Electric engine was designated the F120; the Pratt & Whitney engine was designated the F119. At the same time, USAF requested proposals for a concept definition investigation for ATF.

Boeing, General Dynamics, Grumman, Lockheed, McDonnell Douglas, Northrop, and Rockwell responded to the request and prepared proposals for submittal in mid-June 1983. Just before the deadline, ASD announced a week delay and informed contractors to await further instructions. At the end of June, the companies were asked to add another volume to their proposals to describe stealth-related skills and experience. Any detailed discussion of stealth technologies had evaded the ATF program to this point. The original proposals were limited to thirty pages. The stealth addendum – to be submitted as a separate, highly classified volume – had to fit on five pages.

“Originally, the ATF program did not contain stealth,” explains Al Piccirillo, the director of the ATF System Program Office at the time. Piccirillo is now the manager of the technology division at ANSER, a government consulting firm in Washington, D.C. “People on the program were aware of what was going on in the F-117 and the B-2 programs,” Piccirillo continues. “We would have been really stupid to develop an advanced fighter without using this new technology. Without stealth, I am not sure the Air Force could have justified ATF.”

Including stealth set an unusual security precedent. The security level of the original request for proposals for this phase of the program precluded any details on stealth, a topic that was highly classified in the early 1980s. Companies could claim that low-observable technologies would be considered in a design, but they couldn’t reference any actual experience or techniques in their proposals. Stealth technologies were considered “black.” Such programs did not exist to anyone not cleared on them. The last-minute change to the request for proposals placed the program in both worlds: black and white.

The proposals from most companies for the concept exploration phase showed how they would narrow down their previous approaches for achieving air superiority. This work would lead to the next phase of the program, the demonstration/validation phase, in which they would have to prove their technologies and refine their designs. Lockheed, however, took a radical departure from its high-speed, high-altitude design and started from scratch with an F-117 derivative in its proposal for the concept definition phase.

“Clearly, ATF was going to be superstealth and not a cousin of YF-12 or SR-71,” explains Osborne. “I stopped the YF-12 derivative effort, and we started working on an F-117 derivative for ATF.” The design submitted in the Lockheed proposal looked like a larger and elongated F-117 with some significant differences. It had a high wing rather than low wing and four tails instead of two. The inlets were placed below and behind the leading edge of the wing. The highly faceted airplane weighed around 80,000 pounds and was far from aerodynamic.

“We knew we would have serious problems with the supersonic requirements,” recalls Osborne. “Our design could go supersonic, but it was a real dog of an airplane. With enough power, you can make a brick fly. We did not know how to analyze a curved stealthy shape in those days. The software wasn’t sophisticated enough, and we didn’t have the computational capacity we needed. We had our hands tied by the analytical problems. Lockheed had become convinced that, if we could not analyze a design as a stealthy shape, then it could not be stealthy. We would not break through that barrier until 1984.” Lockheed’s submittal for the concept exploration phase was not received well by the Air Force. The company placed last in the field of seven.

Each of the seven companies that bid on the concept exploration phase of the ATF program, including Lockheed, received a contract for about $1 million. This phase ran from September 1984 through May 1985 when the Air Force received many briefings and thousands of pages of reports that would take the program into the demonstration/validation phase. In the dem/val phase, four winning companies would be given about $100 million each to demonstrate technologies needed to build their ATF. The request for proposals for the dem/val phase was issued in September 1985. The deadline for the proposal was set for that December.

Lockheed’s proposal configuration, called 090P, had a streamlined nose, trapezoidal wing planform with positive sweep on both the leading and trailing edges, and four tail surfaces (two horizontal and two vertical). The large vertical tails were canted outwards. The leading and trailing edge sweep angles of all of the surfaces were aligned at common angles. The design had a wide strake that ran in a straight line from the wing leading edge outboard of the inlets to the point of the nose.

The Lockheed design
After a poor showing in the concept exploration phase, Lockheed had to turn around its ATF program before the next proposal was due.

The company had just lost what became the B-2 with a faceted design to a more aerodynamic flying wing design from Northrop.

Lockheed had also been cut from consideration in the Navy’s Advanced Tactical Aircraft program after entering that competition with a highly faceted design. The Air Force’s response to Lockheed’s concept exploration proposal forced the company to rethink its commitment to faceting for stealth.

“We simply started drawing curved shapes,” recalls Osborne, “even though we could not run the designs through our analytical software models. When we went to curved airplanes, we began to get more acceptable supersonic and maneuver performance. Instead of relying on software models, we built curved shapes and tested them on the company’s radar range. The curved shapes performed quite well in the radar tests.”

The Lockheed configuration quickly progressed from faceted to smooth. The configuration just preceding the company’s final dem/val design, called Configuration 084, was smooth except for a faceted nose. “We knew how to make a stealthy flat radome,” recalls Osborne, “but we didn’t know until early 1985 how to make a stealthy curved radome. We started drawing them in late 1984, before we knew how to analyze them.”

The proposal configuration, called 090P, had a streamlined nose, trapezoidal wing planform with positive sweep on both the leading and trailing edges, and four tail surfaces (two horizontal and two vertical). The large vertical tails were canted outwards. The leading and trailing edge sweep angles of all of the surfaces were aligned at common angles. The design had a wide strake that ran in a straight line from the wing leading edge outboard of the inlets to the point of the nose.

One requirement that drove all of the ATF designs was a wide field of regard for sensors. The requirement called for a 120-degree radar field of regard on each side of the nose. A forward-looking infrared search and track capability was also desired. Lockheed approached the field-of-regard requirement for the radar with three radar arrays placed in the nose of the aircraft (one facing forward and two facing sideways). Each wing root carried an infrared search and track system that operated through faceted windows.

The aircraft carried six air-to-air missiles in a rotary missile launcher. The launcher was loaded away from the aircraft. (Lockheed also designed a version of the launcher that could be used independently for airfield defense.) When closed, the bottom of the launcher became the lower skin of the aircraft.

Lockheed built a large-scale model of a curved configuration to test on the company’s radar range. The data from these tests went into the company’s proposal for the dem/val phase. “The real question USAF had was whether Lockheed could design a curved stealthy airplane,” Osborne explains. “We showed them with the range model that we could do curves.” Lockheed’s biggest advantages going into the dem/val competition were its revamped approach and its vast stealth experience. Lockheed had also earned a fine reputation for rapid-prototyping in a variety of programs, most recently in the Have Blue program.

The Boeing design
The Boeing concept was a larger aircraft than the designs submitted by Lockheed or General Dynamics.

The company retained the higher operating speeds assumed in its previous work.

The most notable features of the design were twin vertical tails located well aft on the fuselage behind a trapezoid planform wing. The vertical tails were sized to provide the same vertical and horizontal control power as four tails. “Our designers argued most over two tails versus four tails,” Hardy says. “The whole Boeing company got involved in the argument. We had special teams set up to study the problem. Two tails won out. Our higher operating speeds led to a longer airplane, which produced a longer moment arm for the tails. So we didn’t need as many tail surfaces. We thought we could meet all the requirements with two tails, which gave our design a lower signature and a lighter weight.”

Boeing designers focused on the weapon bay and essentially designed the airplane around it. Wind tunnel results, especially those related to flight at high angles of attack, affected the arrangement, size, cant angles, and placement of the tails. The design used a single chin inlet with an internal splitter to feed the two engines. The inlet had an internal variable ramp (combined with the splitter) to reach its higher design speeds. Boeing designers moved the nose landing gear aft of the inlet in one of the later design changes. The company had been working on advanced composite materials for USAF labs and in some classified programs in the 1970s and 1980s. As a result, the design used a unique thermoplastic manufacturing process and material for the wing.

The Boeing design carried its air-to-air weapons internally, though larger air-to-ground weapons were carried partially submerged. Heat-seeking missiles were carried in separate bays placed forward in the fuselage. The weapon bay concept relied on quick-change pallets to position the munitions so they could be loaded quickly to meet quick turnaround requirements. The Boeing design had three radars in the nose, one large forward-facing array and two smaller side facing arrays, to meet the 120-degree field of view requirement. Two infrared search and track sensors were placed near the nose as well.

Boeing had done well in the previous phase of the program, placing high in the field of seven. Its design was also well developed and wind tunnel tested. Further, the company had extensive experience in integrating avionics. This experience, which dated back to the AWACS program, had been more recently refined in the B-2 bomber program. The company also had an impressive production capability developed in its commercial airline work.

The General Dynamics design
The General Dynamics design for the dem/val phase evolved from a variety of inputs. During the previous program phase, the company had focused on three separate families of aircraft: conventional, all-wing, and semi-tailless (denoted in the configuration studies by C, W, and T, respectively). The conventional family derived from the Model 21 designs of the previous studies. The all-wing family strove to carry Sneaky Pete’s minimum observables into the supersonic regime. The semi-tailless family, which had a single vertical tail, fell in between these two extremes. After a series of internal design competitions and trades, the company went with the semi-tailless approach.

The wing planform and airfoil design were chosen to minimize weight while providing the maximum turn capability and supersonic cruise. The single vertical tail, however, presented problems in achieving a totally stealthy design. General Dynamics ran many wind tunnel tests to find a location and shape for twin canted vertical tails on the T configuration. The vortex flow off the forebody and delta wing produced unstable pitching moments when it interacted with twin tails. Without horizontal tails, the aircraft did not have enough pitch authority to counteract these moments. A single vertical tail and no horizontal tails was finally identified as the best overall approach to the design despite the degradation of radar cross section in the side sector. The proposal configuration was designated T-330.

General Dynamics took a unique approach to the sensor requirements, using two radar arrays and one infrared search and track sensor. (Boeing and Lockheed had each used three arrays and two IRST sensors.) One IRST sensor was placed in the nose of the aircraft and the two radar arrays were located aft of the cockpit. The radar beam from each array could be steered sixty degrees from the face of the array, allowing each radar to cover the area from straight ahead to 120 degrees aft. The arrays were located just above the engine inlets.

The General Dynamics configuration achieved a high state of detailed design. The company built a full-scale mockup and was finalizing a half-size pole model for testing the design’s radar cross section. Preliminary structural designs were developed, along with locations for manufacturing breaks to allow the aircraft to be divided among potential partners. General Dynamics had done well in the concept exploration phase of the program, placing very high in the field of seven. Among General Dynamics strengths were its extensive experience in fighter design and manufacturing gained in the F-16 program. The company also had experience with rapid prototyping: the YF-16 was an unsurpassed program in this respect.

Amended request
A few months before the proposals for the dem/val phase of the program were to be submitted, the Air Force amended its proposal request. The change significantly increased the importance of stealth in the design. Lockheed, with a stealthy configuration derived from the F-117, made no modifications to its design as a result of the new requirements. Boeing made some slight modifications to the design of their inlet to address the increased stealth requirements. The company was, however, satisfied that its twin-tail design would meet the stealth requirements.

The upgraded requirements forced engineers at General Dynamics to again reconsider twin tails in a variety of locations, including out on pods on the wing. The trailing edge of the wing and the control surfaces were cut into chevrons aligned with the leading edge, giving the wing a bat-like look. In the end, no acceptable location for the twin tails was found, and the design was submitted with a single centerline tail and a serrated trailing edge. The new final configuration was labeled T-333.

Another delay
As it had done with the proposal for the previous phase, the Air Force delayed the submittal date for the dem/val proposals. This time the deadline was put off for prototyping. The amendment required contractors to build two prototypes: one with the F119 and the other with the F120 engine. The last-minute change resulted from a reaction to a report released in the early-1980s by a congressional commission headed by electronics-industry pioneer David Packard, who had been asked to look at reforms in Pentagon acquisition practices. The report, influenced by the recent success of the F-16 program, favored prototyping for new military aircraft.

“Initially, the proposal request did not contain prototypes,” Piccirillo explains. “ATF was patterned on the F-15 program, which did not have prototypes. The Air Force has gone back and forth over fifty years on the value of prototyping. In the 1960s, through the F-15 program, we did not prototype. We performed studies, ground testing, and wind tunnel testing and went right into full-scale development work. We built test airplanes, but they were very close to the production configuration. After the amendment, the dem/val phase of the program called for best-effort concept demonstrators. We left it up to the contractors to decide how they would demonstrate the critical technologies behind their concepts for an ATF. One of the most critical was shaping for supersonic flight and low observables.”

“We had actually finished the proposal and were within a few days of turning it in when we got Modification Request MR-006 to the RFP,” recalls Moran. “Instead of approximately $100 million contracts for four winners, the Air Force added flying prototypes to the program and would award only two contracts of about $700 million each. We were directed to write one more proposal volume describing how we would design, build, and test two flying prototypes-one with the Pratt & Whitney F119 engines and the other with the General Electric F120 engines. We were also required to build a ground-based avionics test lab, and we could offer a flying avionics testbed if we thought that was desirable.” The companies were given sixty extra days to modify their proposals.

About the time the request for proposals for the dem/val phase was amended in late 1985, USAF sent out a letter to the competing companies to encourage teaming. “The amended proposal request had a cover letter that encouraged teaming,” recalls Piccirillo. “The Air Force encouraged teaming because it wanted the best resources from industry to be brought to bear on the program. The program was going to be expensive and big. The more commitment we had from industry, the more likely the program was to succeed.”

A complicated dance among the contractors began immediately to see who wanted to partner with whom. Representatives from Boeing, General Dynamics, and Lockheed signed a teaming agreement in June 1986. Northrop and McDonnell Douglas announced their team two months later. The two remaining companies did not team.

The teaming agreement among Boeing, General Dynamics, and Lockheed called for the winning company to be the team lead. The teaming deal was done “blind,” none of the participants got to see the other contenders’ aircraft or program plans before the contract was awarded. Boeing, General Dynamics, and Lockheed each identified a group of twenty high-level managers who would travel to the winning company’s site the day after the award was announced.

The $700 million announcement came on Oct. 31, 1986, naming Lockheed and Northrop as the top two contenders. Representatives from Boeing and General Dynamics met their Lockheed counterparts for the first time on Nov. 2 as partners at Lockheed’s Skunk Works in Burbank, Calif. The team would be competing against Northrop and McDonnell Douglas for the Air Force’s next-generation fighter.

Secretary Aldridge’s Halloween announcement meant that Lockheed and Northrop had each won $691 million contracts to proceed to the demonstration/ validation phase of the Advanced Tactical Fighter program.

The newly formed teams now led by the two winning companies would build two flying prototypes each one with Pratt & Whitney engines and the other with General Electric engines. Each team would also build ground-based avionics prototypes and a flying laboratory to demonstrate the avionics. The dem/val phase would determine which team would enter the next phase of the program and, ultimately, which team would build production versions of the ATF.

After the dem/val award announcement, the Advanced Tactical Fighter program returned to its stealthy status. Very little information about the project appeared in public until the prototypes were unveiled almost four years later in August 1990. Though concealed in quiet secrecy, work during the intervening four years was anything but quiet for those involved, as the ATF competition intensified.

Show and tell
The dem/val award was announced on a Friday afternoon. The next Monday morning, representatives from Lockheed, Boeing, and General Dynamics met for the first time as a team at Lockheed facilities in Burbank, California. About 100 engineers and managers crowded into a large high-security conference room in Building 360, where representatives from each of the three companies were allotted two hours to present their proposed ATF concept. Lockheed went first. Boeing presented after lunch and was followed by General Dynamics late in the afternoon.

The all-day show-and-tell was unprecedented for everyone in attendance. Never before had they shared everything they knew about a program with an audience considered to be the competition only a week before. Though the three companies had a teaming agreement, they had not exchanged information. Doing so was not possible given the security clearances involved and the short time between the agreement signing and the contract award. Everyone saw everything at once.

“That Monday was the most fascinating day I ever spent in the aircraft business,” remembers Randy Kent, the ATF program director for General Dynamics from 1985 to 1991. “Typically, we never know what other teams have done for months, if ever, after a contract is awarded. For ATF, however, each company made the same presentation at Burbank it made to the Air Force. We all put our models, layouts, and drawings on the table. Everyone received detailed views of what everyone else had done to that point in the program. The experience was amazing.”

“The diversity of the proposals was surprising,” adds Gerry Murff, the chief engineer for General Dynamics. “The depth and capabilities were impressive. Lockheed’s strength with its signature capability became apparent. Lockheed engineers knew how to design aircraft using a fixed set of angles, and they understood all of the supporting detail design technology. Their operations analysis showed why stealth was relevant. These factors were decisive to their win. They had an appreciation for the handling qualities required for the ATF, which led to their four-tail design. They also had a plan a very good plan.”

The importance of that plan was reinforced in another, equally astonishing meeting that happened the day after the initial team meeting when Rick Abell, the Air Force’s lead technical evaluator, briefed the three companies on how their dem/val proposals were evaluated. Abell used the same evaluation charts the F-22 system program office had used when selecting the two winners. He went through about seventy strengths and about thirty weaknesses for each of the three proposals.

“This was the only time in my career that I saw an official government evaluation of what we and two of our strongest competitors had submitted for a competition,” remembers Sherm Mullin, the program director and team leader for Lockheed. “And the evaluation came from an authoritative person, not through rumor. The system engineering volume of the proposals weighed very heavily in the selection. Advocating a point design, a single answer, hurt most of the proposals. Air Force officials made it clear in late 1986 that they wanted to see trade studies. They wanted every requirement challenged. They entertained alternate approaches to almost anything.”

Lockheed’s plan was in fact a key to its win. “Everybody met the requirements in our evaluation of the proposals for the dem/val phase,” explains Abell, who became the Air Force’s chief engineer for the ATF program. “The biggest single important point in our evaluation was risk reduction for the production configuration, what we called the preferred system concept. We weren’t looking at what the prototype would do or what the avionics would do or how they would perform. We wanted a program that was laid out to reduce the risk and develop the technology sufficiently enough so that when we started the next phase it would be a lower risk program. We didn’t spend a lot of time looking at what the proposed prototype airplanes would do in terms of performance.”


Defining the team
Soon after the team met in California, workers split into two basic groups. One group addressed interfaces, costs, and teaming issues for the prototype and, later in the dem/val phase, for the preferred system concept. After splitting the work, the companies went into a difficult and complex process of validating the dollar amounts that each team member associated with its portion of the work. The calculations were complicated by many factors, including differing labor rates and unique estimating processes.

The other group focused on those seven categories of task assignments as apportioned in the teaming agreement, which established the basic relationship among the three contractors. The fifty-page document spelled out the roles and responsibilities of the team leader; the division and assignment of work among the team members; the preparation of future proposals; the handling of proprietary information and patents; the procedures for resolving disputes, cost sharing, and cost reporting; coordination of publicity; and termination procedures for the agreement itself.

Each company had provided its own list of task assignments should it win the award for the dem/val phase (and, therefore, become the team lead). These assignments fell into six categories: weapon system integration, airframe design/systems, avionics, system test, manufacturing, and supportability. A seventh category, systems engineering, was added after the dem/val contract was awarded.

As the team leader, Lockheed’s plan for dividing the work carried the most weight. Accordingly, Lockheed claimed the forward fuselage and nose landing gear, all the specially treated edges and low-observable antenna integration, cockpit, controls and displays, core processing for the avionic systems, final assembly and checkout of the airplane, and leadership of flight test.

Boeing got the aft fuselage and the wings, fire protection system, life support system, auxiliary power system, arresting gear, radar, infrared search and track system, mission software, flying avionics laboratory, and the biggest share and leadership of the training system.

General Dynamics got the mid-fuselage and all of the subsystems in it, main landing gear, horizontal and vertical tails, flight controls, communication-navigation-identification system, electronic warfare system, inertial reference system, stores management system, the infrared portion of low observables, and the biggest share and leadership of the support system. This basic division of work, with some minor revisions, exists to this day on the F-22 program.

The teaming agreement also established the proposed design from the winning party as the starting point for the dem/val phase. The agreement carried the following stipulation: “Such a proposal may be revised to incorporate aspects of the other parties’ proposals for the dem/val phase as such incorporation may be requested by the Air Force or agreed by the parties and concurred in by the Air Force.” In simple terms, the team was encouraged to draw from the strengths of all three individual companies to come up with a winning ATF design.

“After the contract was awarded, it took us two years to figure out the design,” explains Dick Hardy, the ATF program director for Boeing. “We decided that we shouldn’t build a demonstrator until we knew what the production configuration looked like. The Northrop team ran off and made a demonstrator and then tried to figure out the production configuration. When it came time for the downselect for the next program phase, they had to change the location of their weapon bays and a whole bunch of other things. We wanted the data we derived from building and testing the demonstrator to apply to the production version. That relationship was the whole purpose of flying the demonstrators—to produce data useful to the final design.”

When the team first met in Burbank, the winning Lockheed design was represented by an internal arrangement and external line drawings labeled Configuration 090P and a detailed three-view drawing called Configuration 090P/092. Configuration 090P designated the Lockheed ATF proposal for the dem/val phase. Configuration 090P/092 contained changes that occurred during the months the proposals were being evaluated by the Air Force. The differences included a revised inlet and small changes in wing and tail sweep angles. The vertical tails were also moved farther outboard and the chine was narrowed slightly.

The transformation of 090P into Configuration 1132, what is better known as the F-22 prototype or YF-22, involved some of the most concentrated work in the history of aircraft design. The transformation got off to a strained start as the team members sized up their relative strengths and weaknesses and argued for and against a variety of design features. “The period was intense,” says Paul Martin, Lockheed’s deputy chief engineer for technology and design during the period. “We spent a lot of time convincing each other what great he-men engineers we all were.”

The posturing was fed by the sheer amount of material available to scrutinize as all three companies placed their work on the table. Every one of the designs proposed by the three teaming companies had its share of problems and advantages. As the official starting point, however, Lockheed’s design was open to the most scrutiny and criticism.

“After studying the design of Configuration 090P,” recalls Murff, “we soon realized that the airplane would not fly. Its huge forward glove made the design uncontrollable in the pitch axis. The internal arrangement would not go together. The large rotary weapon bay pushed engines and inlets outward, which produced an excessive amount of wave drag. And the rear-retracting landing gear design was not suited for a fighter.”

“After the General Dynamics team had been out in Burbank for about two weeks, they sent home a set of drawings of the winning design,” remembers Kevin Renshaw, the configuration design lead for General Dynamics. “The first task for the engineers in Fort Worth was to put the aircraft drawings into the computer to provide a base for analysis. The immature status of the Lockheed design became immediately apparent. The plan view, profile view, and sections on the drawings had only a rough relationship to each other. After analyzing the design, it became obvious that the aerodynamic and weights data in the proposal had been ‘goal’ levels with little actual relationship to the drawings. The design turned out to be a series of unconnected sections drawn around individual portions of the aircraft’s subsystems. Lockheed had a concept for an aircraft, not a point design. However, that approach won the competition.”

The 090P design may very well have been less defined than the design submittals from Boeing and General Dynamics. But in Lockheed’s defense, the company had gotten off to a relatively late start on its design because it had changed its basic approach just after the last phase of the ATF program. That change catapulted Lockheed from last to first place in the evaluation of the dem/val proposals. Conversely, Boeing and General Dynamics had placed too much value on point designs when the Air Force evaluators were more interested in a solid plan for reducing risk for the dem/val phase.

“General Dynamics historically has focused on the configuration,” says Kent. “We went into enormous detail on the structural design, the aerodynamics, and the wind tunnel testing to present a creditable story that our design would do what we said it would do. We understood that the Air Force evaluators were as interested in detailed dem/val planning as they were in the configuration. In our view, however, we believed that valid trade studies could be made only from a solid configuration starting point. It turned out that we probably misread the Air Force’s emphasis. They were more interested in the details of the dem/val planning than in the validity of the baseline configuration for the trade studies.”

“Configuration 090P was not a point design by plan,” explains Mullin, “not by accident. Our proposal supported what the customer wanted to do and what we needed to do to win the competition. If the customer wants two years of good system engineering and not a lot of point design, then it’s not a bad idea to propose two years of system engineering and not a point design.

“Our team had a big problem at the beginning with system engineering,” Mullin continues. “A lot of the engineers wanted to design an airplane to the stated requirements instead of performing trade studies and allowing Air Force officials to adjust the requirements where they felt the impact was worth the change. One of the main themes of 1987 was to get everyone adjusted to the fact that none of the requirements should be viewed as final. Trade studies and risk-reduction were more important.”

“Everyone was a little defensive of his design at first,” recalls Hardy. “In retrospect, the team did a great job coming together and reaching an agreement. Mullin did a great job. He kept everyone focused on the design and on the program. He kept arguments from getting personal, and he always focused on what was best for the program.” Mullin led the ATF program for the Lockheed-Boeing-General Dynamics team through the dem/val phase. He had the unenviable task of creating a cohesive group from three divergent corporate cultures that had each spent years evolving its own unique ATF design.

“I think I was reasonably fair with all,” Mullin explains. “On an average week, I had just as many Lockheed engineers upset with me as I did Boeing and General Dynamics engineers. Everyone believed that his own way of approaching a design was practically biblical. I had to convince people from Boeing and General Dynamics that I was playing fair and square and that I did not have a Lockheed bias. That was not easy. However, I developed a very productive relationship with Hardy, Kent, and with many others. We became a cohesive team, and that was a key factor in winning the competition. We did not change each other’s organizational culture very much. We simply learned to live with each other and work together. Al Pruden, as the system engineering team director, had the tough job of getting this done right.”

“The three company program managers Mullin, Hardy, and Kent formed a complementary set,” observes Bill Moran, who attended the initial meetings in 1986 and still works on the program. “Mullin is excitable, ebullient, and fascinated with technology. Hardy is fiscally sensitive, hardheaded, and laconic with a droll sense of humor. He was the perfect counter to Mullin’s excitability. Kent is reserved, intellectual, and dedicated to producing a real airplane that excels in flight test and in eventual operations. He kept the other two pointed at the ultimate objective of the program: to produce an operational weapon system. Watching these leaders interact was always exciting and instructive. Seeing how well they got along, even when they disagreed, was good for the whole team. As with any other group of several thousand people, some pairings worked and some didn’t. But upper management’s lead established the right tone for success. I haven’t been on any other team, but I’ve heard stories about other teams that make it clear that a team won’t work unless it has strong examples at the top.

“Jim Fain, who commanded the ATF system program office during dem/val, also deserves a big share of the credit for making this program and this team work,” Moran adds. “He turned out to be even smarter than most of us originally perceived. And he was a fearless innovator. When he joined the program as a colonel at the beginning of dem/val, he announced that he was never going to make general so he was going to run this program the way he thought it should be run. Fain eventually was promoted to brigadier general in the middle of the dem/val phase and retired as a lieutenant general. He personally forced the team to work on both the Air Force side and the contractor side. He used competition between the two teams rather than program direction to get what he wanted. With the help of Rick Abell and Col. Tom Bucher, he introduced many of the acquisition reforms that were made across the Department of Defense several years later.”

The design process
Mullin, Hardy, and Kent became the ultimate arbitrators in a design process that required new levels of tact, efficiency, and arbitration for the Lockheed-Boeing-General Dynamics team.

“The working troops would communicate disagreements through their company chief engineer, who worked for one of the three program directors,” explains Kent. “If they could not come to a resolution, they had to work first through their respective chief engineers. If they still couldn’t decide, the issue was kicked up to the Mullin-Hardy-Kent level. No matter how good an engineer was, he had to learn how to sit down, let the other person speak, treat him with a certain measure of respect, and not ridicule him. We removed people who could not get along. Some of the best technical contributors had the hardest time living with their counterparts. We all came to the table with traditional ways of doing things and with different backgrounds from different airplanes. I began to sense that we were becoming a more cohesive team when the aerodynamicists from all three companies began voting against the structural engineers from all three companies.”

The design process was influenced by regular contact with USAF representatives and by more formal USAF reviews. “The Air Force controlled the evolution of the ATF with a draft specification that spelled out the initial requirements for the aircraft,” Abell explains. “We adjusted the system specifications once a year during the dem/val program as studies came in and alternatives were evaluated. In our system requirements reviews, we took the operations analyses and the results of tests and other information and determined if we were asking for too much or too little. We adjusted the specification accordingly. If members of one team had trouble making a requirement and we didn’t change it, they soon appreciated that maybe the other team was meeting the requirement. When a requirement changed, both teams knew it was hard for everyone to meet.”

For example, the Air Force initially required that eight missiles be carried internally in the main weapon bay. “One team thought it could be done,” says Abell, “but it didn’t know for sure. Consequently, we didn’t change the requirement to six missiles until both teams decided that eight could not be carried effectively. In another example, short field requirements were changed to eliminate thrust reversers because the benefit was not worth the price we had to pay for the capability.”

“The evolution of the production configuration involved an annual thrash on system requirements with the ATF System Program Office and with Tactical Air Command,” adds Moran. “Each year of design work produced refined weight, cost, performance, and effectiveness estimates. These estimates led to changes in the requirements, which required new design work, which led to new changes in requirements, and so on.”

Air Force involvement in the design process during dem/val was, therefore, somewhat indirect. “Our biggest influence on the program was allowing the companies to explore and develop rationale and reasons for specific systems,” says Abell. “Our involvement in the technical aspects was more to understand the design concepts and the approaches being taken. We were always asking, ‘Why are you doing that?’ Not so much suggesting, ‘Why don’t you do this?’ We were more interested in the processes and the reasoning behind the designs than in any particular detail. If anyone on my team said ‘Do this,’ I shot him. We could not give direction.”

Design challenges
The basic challenge of the ATF design was to pack stealth, supercruise, highly integrated avionics, and agility into an airplane with an operating range that bettered the F-15, the aircraft it was to replace. The F-22 also had to have twice the reliability and half the support requirements of the F-15.

“One problem we typically face when trying to stuff everything inside an airplane is that everything wants to be at the center of gravity,” Hardy explains. “The weapons want to be at the center of gravity so that when they drop, the airplane doesn’t change its stability modes. The main landing gear wants to be right behind the center of gravity so the airplane doesn’t fall on its tail and so it can rotate fairly easily for takeoffs. The fuel volume wants to be at the center of gravity, so the center of gravity doesn’t shift as the fuel tanks empty. Having the center of gravity move as fuel burns reduces stability and control. We also had to hide the engine face for stealth reasons. So, these huge ducts had to run right through the middle of real estate that we wanted to use for everything else. The design complexities result in specialized groups of engineers arguing for space in the airplane. That was the basic situation from 1986 through 1988.”

As the design progressed, weight became the most difficult design challenge. “We were never able to design an airplane with a 50,000-pound takeoff gross weight that came close to meeting the Air Force requirements,” Mullin admits. “After two years of hard work, we convinced the Air Force that it was not possible. The weight requirement was changed.

“Getting the right inlet design was very difficult as well,” Mullin continues. “The aerodynamic, structural, low observable, and producibility requirements conflicted with each other to a significant degree. We took about three years and performed a lot of analyses and wind tunnel testing before we achieved a fully acceptable inlet design. In the end, the performance was excellent.”

The first major change in the Lockheed-Boeing-General Dynamics configuration came in February 1987, when a more space-efficient flat weapon bay was substituted for the rotary weapon bay found on 090P. The change marks the transition to Configuration 095. The deletion of the rotary weapon bay allowed the engines to be moved closer together, which reduced wave drag.

The large hooded strake was narrowed to reduce the planform area ahead of the airplane’s center of gravity and to push the vortex generated by the strake farther back. These changes were essential for controlled flight at higher angles of attack. Engineers also worked to repackage the forebody to reduce cross-sectional area to improve high alpha capability and to reduce wave drag and weight. The engine inlets were also redesigned. Configuration 095, however, retained the trapezoidal wing and tail design found on 090P.

At about this time, the configuration evolution was split into two families. The first family, denoted with a 1000 prefix, represented the prototype design (called the prototype air vehicle). The second family, with a 500 prefix, represented the preferred system concept—the design that would be carried into the proposal and the next phase of the program. Configuration 095 thus became 1095 (prototype) and 595 (production).

Minor iterations within a configuration were represented with dashed numbers. In July 1987, for example, the preferred system concept was Configuration 595-6. As the dem/val phase progressed, these two design families grew further removed. The prototype design was frozen so the design could be built and flight-tested during dem/val. The preferred system concept evolved toward a production design for the next program phase (the full-scale development phase, a term later changed to engineering and manufacturing development phase).

The first requirements review with the Air Force came in May 1987 and lasted about a week. “The first review was a communication session to make sure that we were on the right track,” Mullin recalls. “I think we all came out of the first review feeling good about our design. However, at that point, optimism still reigned supreme. No one was ready to admit that our ATF was not going to be a 50,000-pound airplane. We had not done enough of a detailed design and weight analysis to get a good set of weights.”

The great configuration chase
The team got those weights in late June 1987, just before a three-company executive meeting in Fort Worth on July 10. “We were almost 9,000 pounds off on gross weight and $5 million off on unit cost,” recalls Kent, “but we were still making all the maneuver parameters. We had a pretty good airplane except for the weight and cost problems. We wanted to relax the fuel load at max g requirement. We wanted to let gross weight be a fall-out from meeting the other requirements, delete a couple of the missions, and work on cost by simplifying the avionics requirements.”

When those compromises were not forthcoming, the team decided to step back and open up the design to more fundamental changes. “After a bloody debate, we agreed to trash the current design and start over,” says Mullin. “Over that weekend, we brought in a new director of design engineering, Dick Cantrell, flew in people, and started a ninety-day fire drill. Work started on July 13. That date marks one of the most creative periods of conceptual design for any fighter aircraft. We looked at different inlets, different wings, and different tail combinations. One configuration had two big butterfly tails and looked somewhat like the F-117, though people did not know that since the F-117 was still highly classified. The configuration search was wide open, but the biggest single change that resulted from it was to go with diamond-shaped wings.”

The concentrated configuration search began with a slew of possible designs. The search complicates the numbering scheme considerably, as diamond wings, twin tails (two tails instead of four), various inlet shapes, and various forebody shapes were all considered and reconsidered simultaneously in the summer of 1987.

Configuration 595-7, with trapezoidal wings and four tails, was the starting point for these studies. Configuration 608A, labeled the baseline equivalent, represented a similar shape with six instead of eight missiles in the main weapon bay (all the design alternatives assumed this loading).

Configuration 608 had a trapezoidal wing and twin tails. Configuration 607-0 introduced the diamond-shaped wing similar to the General Dynamics proposed ATF, but with four tails instead of one. Configuration 607-11A had a diamond wing and twin tails. Configuration 611A had a diamond wing, twin tails, and the large chine reminiscent of 090P. Configuration 609A looked a lot like the Boeing proposed ATF with its trapezoidal wing, twin tails, and single chin inlet.

Configuration 610A was a Boeing/Lockheed hybrid with trapezoidal wings, twin tails, and twin side inlets.

Computer-aided design proved critical to completing these configuration studies. The ATF effort was one of the first fighters to be designed from the beginning with computers. At first, the team used CADAM, a relatively fast two-dimensional drafting package.

“CADAM was better suited to designing detailed parts, not iterating design concepts,” explains John Hoffschwelle, a configuration designer for General Dynamics who spent months in Burbank working the ATF design. “We quickly realized that we needed ACAD and our Perq computers to speed up the design process. We had them flown out from Fort Worth and approved by Lockheed security, which was no small feat. About six of us Fort Worth engineers went to work in a small office, or big closet. ACAD allowed us to look at more ideas and at each idea in greater detail.”

ACAD, a three-dimensional software package developed by General Dynamics, was used for conceptual and preliminary design. (A much improved version is in use today.) The team also used a process that linked ACAD with CATIA, a high-fidelity three-dimensional software package developed by Dassault. CATIA can output data directly to numerically controlled machining equipment. The ATF designers would do a first-generation lines database on ACAD so they could be iterated quickly. These files would then be translated into a master CATIA database.

“We also had datalinks so that design data and other data could be moved digitally among all three sites,” Mullin adds. “This network, set up early in the program in 1987, was created in the days of mainframes, not workstations. The datalinks were secure and encrypted. An engineer in Fort Worth could share design-related data at Burbank or Seattle.” By mid-August 1987, the configuration choices had been narrowed to five, represented by Configurations 595-7 (baseline with trapezoidal wings, four tails, and eight missiles); 612 (baseline equivalent with six missiles); 613 (trapezoidal wings and twin tails); 614 (diamond wings with four tails); and 615 (diamond wings, twin tails, and twin side inlets). By late August, the diamond-wing four-tail Configuration 614 won out.

“The fundamental reason for going to a diamond wing was that it provided the lightest configuration and gave us the best structural efficiency and all the control power we needed for maneuvering,” Mullin explains. “The biggest consideration was its light weight. Weight drove the decision.”

“A diamond wing has more square feet of surface area, but is more structurally efficient,” adds Renshaw. “The longer root chord provides a more distributed load path through the fuselage. Multiple bulkheads carry the bending loads. The design provides more opportunity to space the bulkheads around the internal equipment. It also provides more fuel volume.”

“The structural engineers wanted a diamond wing because it provides a larger root chord, which carries bending moments better,” Hardy notes. “The aerodynamicists wanted a trapezoidal wing because it provides more aspect ratio, which is good for aerodynamics. Dick Heppe, the president of Lockheed California Company, made the final decision, and he was right. The aerodynamics were not all that different, but the structure and weights were significantly better. So we went to a diamond shape. The big root chord, though, moved the tails back. Eventually we even had to notch the wing for the front of the tails. If the tails moved farther back, they would fall off the airplane.”

The tail chase
Once the wings were set with Configuration 614, subsequent configurations dealt with the tail arrangement. “We spent a lot of wind tunnel time looking at the tails,” recalls Lou Bangert, the chief engineer for engine integration from Lockheed. “From late 1987 to early 1988, we were engaged in what we called ‘the great tail chase.’ We knew we would have four tails, but where they would go was a big deal. A small change in location often made a huge difference. We had to look at performance effects, stealth effects, stability and control, and drag at the same time. The tail arrangement and aft end design were important design considerations for all of these effects.”

Wind tunnel results showed an ultra-sensitive relationship between the placement of the vertical tails and the design of the forward fuselage. The interactions could not be predicted accurately by analysis or by computational fluid dynamics. The airflow over the forebody at certain angles of attack affects the control power exerted by the twin rudders on the vertical tails. Getting the airflow right was critical.

The cant and sweep angles of the vertical tails could not be altered too much because such changes increased radar signature. In finding a suitable arrangement, the control system designers were constrained by the radar signature requirements to moving the tail locations laterally or longitudinally and to shrinking or enlarging them while holding the shape essentially constant. By the end of the dem/val phase, the team had accumulated around 20,000 hours in the wind tunnel. A lot of this time was devoted to tail placement studies.

Configuration 614-6, with trapezoidal horizontal and vertical tails, represented the starting point for the tail chase in December 1987. After many intervening configurations, the vertical tails had evolved to a diamond shape by February 1988 in Configuration 630. The wing area was also reduced in Configuration 630. The size of each vertical tail increased by seven square feet in Configuration 631. The rudder size was also increased slightly and the cant angle of the verticals went from thirty degrees to twenty-eight degrees. The prototype design was frozen at this shape (Configuration 1131) in March 1988.

The prototype design was unfrozen at the last minute in May 1988 after the Air Force eliminated the requirement for thrust reversing for short-field operations. The change allowed the team to alter the external mold lines on the aft fuselage and nozzles in the area around the thrust reversers. The trimmed aft end reduced drag significantly. “We never had an airplane with the right supersonic drag until May,” Mullin explains. “We scared the Air Force when we unfroze the prototype design at that late date. The supersonic drag was still too high to supercruise. A team led by Ed Glasgow, our chief flight sciences engineer, redesigned the forebody and aftbody. Suddenly we had acceptable supersonic drag levels that ensured that the airplane would supercruise.”

The final design freeze for the prototype occurred at Configuration 1132 in May 1988. Besides the reshaped forebody and trimmed aft section, the horizontal tails also changed from trapezoidal to diamond in the transition from Configuration 1131 to 1132.

The prototypes
The first drawings were formally released for YF-22 production on 1 April 1988. Construction of the first of two YF-22s began in Fort Worth with the rough-cutting of the titanium 631 bulkhead of the mid-fuselage on 27 April. The mid-fuselage was built in the main factory in Fort Worth in a secure area at the north end of the final assembly line for the F-16. Production of the forward fuselage began soon after with the nosewheel forward bulkhead at Lockheed facilities in Burbank. The aft section and wings took shape during the same timeframe at Boeing facilities in Seattle, beginning with the flaperon torque arm assembly. Fabricating the prototypes consumed the next two years.

The various sections of the prototypes came together in Palmdale, California. The first mid-fuselage was shipped to the West Coast in a Lockheed C-5A Galaxy on 12 January 1990. Shortly after taking off from Carswell AFB, the big cargo plane was struck by lightning. Undeterred, the mid-fuselage was readied to mate with the forward fuselage later that night. The forward fuselage was brought up from Burbank on a truck a few days earlier. Boeing shipped the aft fuselage down from Seattle on a truck the same day the mid-fuselage arrived.

“The mating of all these major structural components was rapid and as smooth as silk,” Mullin recalls. “The use of CADAM software for the detailed design of both the airplane and the assembly tools at all three locations really paid off.”

The YF-22 was unveiled to the public on Aug. 29, 1990, at Skunk Works facilities in Palmdale, Calif. The prototype took to the air for the first time on Sept. 29 from Palmdale when Lockheed test pilot Dave Furguson flew it to nearby Edwards AFB. The second prototype flew for the first time on Oct. 30. The flight test program that followed was one of the most concentrated efforts in aviation history.

The flight rate ramped up from 13 in October, to 22 in November, to 38 in December. The team accumulated over 90 flight hours in 74 flights as the YF-22s expanded flight envelopes beyond Mach 2, seven g’s, and 60 degrees angle of attack. The flight test program included the live-firing of both AIM-9M Sidewinder and AIM-120 AMRAAM missiles.

Supersonic flight without afterburners, or supercruise, was demonstrated with both the Pratt & Whitney engines and the General Electric engines. “After we finally made our first flight, our flight rate was extremely high,” Kent says. “It had to be since we had less than three months to fly. In the back of our minds was the YF-16 and YF-17 contest for the Air Force’s lightweight fighter program. General Dynamics had a much higher flight rate on the YF-16 than Northrop had on its YF-17. We wanted to repeat that performance for the YF-22 flight test program, and we did. We had a very experienced set of flight test personnel from Lockheed, Boeing, and General Dynamics. We had a lot of experienced field service personnel as well. Lockheed’s Dick Abrams did an outstanding job as head of our flight test team at Edwards.”

The flight test program required that the teams demonstrate performance and handling qualities representative of the ATF requirements, including supercruise. Thrust vectoring, while not a requirement, was fully demonstrated through post-stall maneuvering. Low-observable requirements were evaluated on radar ranges with full-scale pole models.

“The Northrop-McDonnell Douglas team got off the ground a month ahead of us and executed a perfectly normal and successful short flight test program,” Moran says. “We showed up late and crammed more into three months of flight testing than had ever been imagined. And we pulled it off without a hitch.”

Production design evolves
The production design continued to evolve after the prototype design was frozen. The dem/val evolution post-Configuration 1132 accounts for most of the external differences between the YF-22 and the F-22. By July 1988, Configuration 632 had evolved into Configuration 634, which had a relocated cockpit, a shortened inlet, and an improved structural arrangement. More importantly, it had a $9 million cap on flyaway cost for avionics. The cost cap on avionics had a profound effect on the program.

“The avionic requirements were growing and growing,” Mullin explains. “Senior USAF officials met in January 1989 and decided on the avionics cap. At the time, we had paper designs of avionic systems that were over $16 million per ship. The cap sent a shockwave through the program, as much through the Air Force as through the two teams. The infrared search and track system bit the dust and so did a lot of other systems, including the side-looking radar apertures. That limit is probably one of the best things that happened to the program. Avionic growth had gotten out of control before that point. The Air Force solved a lot of problems by putting one simple number on the table.”

By the fall of 1989, Configuration 634 had given way to Configuration 637, which had different forebody shaping, a new weapon bay design, and a new systems arrangement. Configuration 638 arrived in early 1990. The leading edge wing sweep was changed from 48 degrees to 42 degrees. The inlet moved farther aft and the cockpit farther forward. The vertical tails became smaller. The wingtip trailing edge was clipped. The overall length of the airplane was shortened from sixty-four to sixty-two feet. A variation of Configuration 638 was the final design submitted in the proposal for the next phase of the program. The proposal was delivered to the Air Force at Wright-Patterson AFB, Ohio, on the last day of 1990.

The win and after
Secretary of the Air Force Donald Rice announced the winner of the ATF program on April 23, 1991. Rice noted that the Lockheed and Pratt & Whitney designs “clearly offered better capability at lower cost, thereby providing the Air Force with a true best value.” The original full-scale development contract called for nine single-seat, two two-seat, and two ground test F-22 aircraft. The subsequent production phase originally included the production of 750 fighters with first delivery scheduled for 2005. Late in dem/val, the total was reduced to 648 aircraft. Subsequent post-Cold War funding and threat analyses have now reduced the planned buy to 339 aircraft. However, the potential for an air-to-ground version of the F-22 to eventually replace the F-15E is favored by the Air Force. Sales to U.S. allies is another long-term possibility.

Since the Air Force did not provide a detailed evaluation of the proposals to the contractors after the selection, the reasoning behind the Lockheed-Boeing-General Dynamics win is open to some interpretation. Sherm Mullin explains it this way: “Our goal was to beat the Northrop-McDonnell Douglas team in every aspect of the competition, not just to focus on some selected areas. We used the team investment, which had grown to $675 million, to implement an across-the-board strategy. We had a balanced design for our production configuration. Our prototype performance was almost exactly as we predicted to the Air Force. Our full-scale pole models showed that radar signature objectives were achievable with low risk. Our approach to integrated avionics was demonstrated in our ground-based and flying laboratories. We prototyped an enormous amount of avionics software in real-time, and it worked well. Reliability, maintainability, and supportability features were designed into the F-22 with an intense focus on self-sufficient operations. And we rigorously complied with every detail in the Air Force’s request for proposals for the engineering and manufacturing development phase.”

The F-22 program begins the transition from development to production this fall with the award of long-lead contracts for the first lots of production aircraft. Though the design currently stands at Configuration 645, the external lines have changed very little from Configuration 638, the design proposed for the engineering and manufacturing development phase in December 1990.

Editor’s note: This article is a two-part series that was originally published in Code One in 1998.