Winner Take All

All the nail biting, second guessing, and sheer engineering brilliance in the battle to build the better Joint Strike Fighter.

EACH MORNING DURING THE SUMMER OF 2000, THE STILLNESS OF THE EVERGLADES was shattered by the thunder of an experimental propulsion system mounted 25 feet up on a test stand at a Pratt & Whitney facility in West Palm Beach, Florida. As the alligators stirred in the swamp, the engine roared away, hour after hour, week after week, while Lockheed Martin engineers watched nervously for signs of trouble in their intricate, classified design. All too frequently, they would shut the unit down as one mechanical problem after another dogged their efforts.

First there were oil leaks. Then, a quarter-inch misalignment of two gears produced tiny metal shavings, which worked their way into the gearbox. Bearings failed and a nut wobbled loose. In most test programs, such failures would rank as little more than minor annoyances that ingenuity and patience would surely overcome. But the continual glitches only added to the highstakes gamble that Lockheed was already taking with its revolutionary new propulsion system: a massive lift fan weighing 1.5 tons. This was the company’s daring solution to one of aviation’s most daunting challenges: getting a supersonic fixed-wing airplane to take off and land vertically.

And in early 2000, the clock was ticking for the Lockheed team. At stake was a contract, worth at least $200 billion, to build the Joint Strike Fighter, a one-size-fits-all attack fighter for the Air Force, Navy, and Marine Corps. JSF program officers had already let Lockheed managers know that their chances of winning against Boeing, their rival in the competition, depended on the success of the lift fan. And Lockheed’s engineers were well aware that for all the brilliance of the lift fan concept, its mechanical complexity would be its Achilles’ heel.

In August 2000, propulsion and controls engineer Scott Winship received an ominous summons from Lockheed Martin’s president, Dain Hancock. “Dain dragged us all into his office,” Winship remembers, “because he knew that if we couldn’t finish our testing on time, the customer was going to pull us from the program. All we’d have to show for it would be this neat simulation, and we’d never get to fly. I’ll never forget, he said, ‘I want to look everybody straight in the eye and ask if you’re going to finish this program.’ My first thought was Well, maybe this is my last day at Lockheed. At the end of the meeting he handed us the scepter and said, ‘I want you to go do this!’ ” For Winship and his team, it was a make-or-break moment. “Thinking back, I bet half the people in the room didn’t believe we could make it. And the rest, like me, who were sticking their necks out, thought Yeah, I think we can.”

Boeing and Lockheed Martin’s epic duel began in two of Air Force Plant 42’s giant hangars, separated by less than a mile of runway and Death Valley desert scrub in Palmdale, California. In one building, the home of the Skunk Works, was Lockheed Martin’s team. Across the runway was a former Rockwell facility taken over by Boeing and filled with topdrawer engineering talent, some of it fresh from McDonnell Douglas after Boeing’s merger with the aerospace giant in 1997. The only visitors allowed inside both facilities were JSF program officers and a crew from the Public Broadcasting System’s NOVA series. (This was the first time TV access had ever been granted to a classified military development program. The NOVA documentary about the triumphs and heartbreaks of the JSF competition, on which this article is based, will air on January 14, 2003.)

The last attempt at building one fighter for both the Navy and the Air Force was in the early 1960s, when General Dynamics produced the F-111. After almost a decade of snarling, the Navy backed out of the deal, and the Air Force ended up with one of its most controversial fighters. Now Boeing and Lockheed faced the same danger: In trying to satisfy all three services, they could end up pleasing no one.

The Air Force demanded an agile and stealthy strike aircraft that would enable it to retire its F-16s. The Navy needed a replacement for its F-14s and F/A-18s sturdy enough to operate from a carrier deck. The Marine Corps was wrestling with the shortcomings of the short-takeoff-and-vertical-landing AV-8B Harrier. The Marines stipulated that, unlike the Harrier, their STOVL version of JSF had to be stealthy, supersonic, and able to bring back a 5,000-pound payload at the end of a mission. “When I took an initial look at the requirements,” recalls Boeing’s chief designer, Dennis Muilenberg, “it worried me. It was by far the most difficult set of requirements I’ve ever seen. It needs to do everything that conventional aircraft do. It needs to fly vertically, carry internal weapons; it also has to be low-signature; and, oh, by the way, it has to be very low cost and be much more supportable than previous aircraft. So yeah, I was worried.”

To overcome their greatest worry—STOVL capability—Muilenberg and his team chose the simplest, cheapest solution that had already been tested by an operational aircraft: the direct lift approach, pioneered by the British Aerospace Harrier. A direct lift system redirects the thrust from the engine through a series of downward-pointing nozzles on takeoff and landing. Muilenberg’s design involved channeling most of the thrust through two main lift nozzles close to the center of gravity, while additional nozzles at the wingtips and tail would help control the airplane’s attitude. Digital flight controls would manage the job of coordinating the hover control, eliminating the tricky handling that had made the Harrier such a nightmare for neophyte pilots.

But other drawbacks of the Harrier approach were not so easy to overcome. The total reliance on the engine for lift in takeoff and landing meant that weight was always a crucial factor. “Boeing was the first to get the cost message,” says Flight International reporter Graham Warwick, “and the simplicity of direct lift gave them a great rationale. But like the Harrier, their plane’s STOVL performance always depended on the engine. They were always asking for more thrust from the engine than Lockheed, and always fighting weight from day one. Though every aircraft test program fights weight, for Boeing it became their most critical factor.”

Lockheed’s solution to STOVL was the lift fan, a groundbreaking design that brought with it different kinds of headaches. The concept dates to 1987, when officials from the U.S. government’s Defense Advanced Research Projects Agency asked Skunk Works engineer Paul Bevilaqua to come up with a way to improve the Harrier’s performance. In his subsequent patent, Bevilaqua sketched out his idea: installing a pair of horizontal, counter-rotating fans that would provide a pillar of air for the airplane to hover and land on, in addition to the vectored thrust from the engine. But what would drive this extra source of lift? Bevilaqua had a “Eureka!” moment when he figured out an efficient way to extract additional power from the engine. This power was transferred to the lift fan by a drive shaft that projected from the front of the engine. The drive shaft had to make a 90-degree turn to the horizontal fan via a clutch and gearbox similar, in principle, to those of an automobile.

Bevilaqua’s back-of-the-envelope calculations suggested that the drive shaft could supply a phenomenal 28,000 horsepower, enough to make the lift fan support nearly half the hovering weight of the airplane. “Several of my colleagues sat up and said ‘Holy smoke!’ ” chief engineer Rick Rezabeck recalls, “ ‘You’re going to have 28,000 shaft horsepower running through the middle of a fighter jet.’ That’s about half the level that the Navy puts through the shaft of a destroyer. So the whole question was: Would it hold itself together and could we make it mechanically and structurally sound enough so it was reliable and added up to a viable jet fighter?”

“We’re dead in the water!” For nearly a year, Boeing engineer George Bible had been experimenting with a novel composite material for the delta wing of the JSF, a project that grew out of a series of Boeing decisions to make sturdy and cost-efficient components for its new fighter.

The concept was a winner: Build the wing as a rugged, one-piece metal structure, sandwiched by two layers of composite—an upper skin and a lower skin. To make the skins more durable, Boeing would embed carbon fibers in an advanced thermoplastic resin. But no one had tried to build a wing skin 30 feet across from a single piece of thermoplastic. Now, as Bible stared at his ultrasound monitor, it was clear that the skin was riddled with bubbles.

The experiment had begun encouragingly enough. Bible’s team had spent weeks laying down sheets of carbon fiber into resin until the wing skin was 90 sheets deep but still less than an inch thick. It was then cured in a massive oven-like autoclave under high pressure, which forced the fibers to blend with the resin. Emerging from the oven, the quality of the first upper skin seemed to bode well for Boeing’s gamble. But the lower skin had a more complex shape, and patches of the material ended up sticking to the mold. One of the advantages of working with thermoplastic is that it can be “re-cooked” if defects show up in the manufacturing. Bible’s team added more release agent—similar to cooking spray—to the mold and tried again. This time the skin didn’t stick but the pressure hoses leaked, and out came the bubble-ridden mess that had distressed Bible.

Bible launched a desperate effort to make the advanced resin pay off: If he cooked the wing yet again, perhaps the bubbles would disappear. For 30 hours the team members held their breath. Gingerly, they peeled away the orange pressure bags—and Bible’s face fell. Patches were still sticking to the mold, and there were wrinkles where the resin had been compressed unevenly. Now Bible felt as if the weight of the whole JSF program was on his back. “If we don’t have a wing skin, we don’t have an airplane,” he said. “We don’t make first flight—it’s pretty much ‘game over.’ ”

As the wing-skin crew struggled, Boeing’s main design team wrestled with its own crisis. The Navy had come back with new demands for performance and weapons-carrying capability. Flight simulators revealed that, with the extra weight on its delta wing, Boeing’s airplane could no longer meet the Navy’s demands. For months, the engineers worked on various fixes. Some sparked protracted debate, notably a design for a novel tail configuration advanced by a former McDonnell Douglas engineer, Ralph Pelikan. A normal four-post fighter tail layout features a twin pair of tail surfaces. The Pelikan tail would replace this conventional layout with a striking two-post layout in which just two angled tail surfaces controlled both pitch and yaw.

In October 1998, top Boeing designers weighed the advantages and penalties of Pelikan’s design. One argued that it offered greater pitch control at high angles of attack. Then the stealth experts pointed out that two tails would have a lower radar signature than four. “We can’t afford to have any question at all over our signature,” argued Fred May. “I vote for the Pelikan tail.”

But another engineer came up with a surprising objection: Despite the fact that the Pelikan tail would eliminate the need for two control surfaces, it might actually end up heavier. The bigger hydraulic pumps and cylinders needed to operate the larger surfaces would end up adding at least 200 pounds to the design. Meanwhile, team leader Dennis Muilenberg was worried about customer perception. He believed that the JSF office viewed Lockheed’s conventional fourpost tail as a lowrisk approach. Should Boeing also go with a tried-and-true design? “On the other hand,” he added, “if we end up looking like we’re the followers and Lockheed’s the leader, it might be a strategically bad thing.”

Eventually it fell to Muilenberg to break the stalemate. Despite earlier doubts, he concluded, “We need to do something to our configuration that will give us an advantage. I think the Pelikan tail does that. We’re going to have to work the hell out of weight, but I can’t imagine anyone better at doing that than the Boeing team.”

But days later, Muilenberg’s team reversed its decision. Fresh analysis suggested that the weight penalty of the Pelikan tail might be more like 800 or 900 pounds, and this and other factors tipped the balance in favor of a conventional four-post tail.

Back in Seattle, another key decision put an end to George Bible’s agony. After the third thermoplastic failure, he was told to abandon what Boeing had hoped would be a competitive edge over Lockheed, and revert to more conventional—thermoset—material. His exhausted crew cooked up the required wing skins without a hitch. “It’s just a good feeling being done with them,” Bible said as he watched them being loaded on to a C-5 Galaxy. “They were quite a pain.” The Galaxy roared into the sky over Seattle and delivered the wing skins, more or less on time and on budget, to the Boeing assembly line at Palmdale.

While Bible was struggling with his unruly resins, Lockheed faced its ultimate trial by fire. In early 1999, the first of five test lift fans was hoisted onto the giant Pratt & Whitney test rig overlooking the Florida swamps. As the engine roared day and night, test data was e-mailed daily back to Palmdale, where the engineers would compare results with the predictions of their flight control simulators.

Although the constant mechanical glitches that plagued the tests were highly visible to the media, they were never the real threat, according to engineer Scott Winship. “I always had faith we could solve those kinds of problems,” he says. “What I didn’t know was whether we would succeed in integrating the flight controls we needed to make this huge fire-breathing beast behave. And while we were having all these mechanical problems, the flight controls testing kept getting delayed and we had still not done the hundreds of hours of tests we needed to write the code that makes the airplane fly. The program was squeezed—we just couldn’t get enough data for our answers. So the whole schedule started slipping.”

"Home sweet home!" exclaimed Boeing chief test pilot Fred Knox as he clambered into the cockpit. It was shortly before 8 a.m. on September 18, 2000, and on the runway at Palmdale, Boeing’s first demonstrator, the X-32A, was on the brink of its maiden flight. This morning, Knox’s mission was to fly the X-32A, with landing gear down, to Edwards Air Force Base, half an hour away, where it would undergo another five months of flight trials.

Knox flipped a switch and the engine roared to life. At the edge of the runway, Boeing engineers cheered. “Very shortly after liftoff,” Knox said, “it was absolutely clear to me that I was flying the airplane we had designed, built, and that I had been simulating for several years.”

Dennis O’Donoghue, a second JSF test pilot, following behind Knox in an F/A-18, was in shock. “Fred was flying at military power with no afterburner,” he said. “But he started climbing like a rocket. It was incredible: He was just gone. I had to use full afterburner, and only caught up with him at 10,000 feet.” Although the X-32A sprang a hydraulic leak and was ordered back to Palmdale, the test program was off to an auspicious start.

A month later, Lockheed caught up as its demonstrator, the X-35A, got off the ground. But the most crucial flight trial confronted Lockheed the following summer. Early on Sunday, June 24, 2001, JSF program manager Rick Baker nervously joined the lift fan’s godfather, Paul Bevilaqua, and its two key problem-solvers, Winship and Rezabeck, at the edge of the Palmdale runway. Former Harrier pilot Simon Hargreaves, a British test pilot, was about to nurse the X-35B STOVL version and its lift fan into the air for the first time.

“At the time, we were just supposed to be doing ‘press-ups,’ where Simon was going up to only five feet,” says Baker. “And he did five feet—we watched the wheels come off the ground and my heart started beating faster. Then he went up 10 feet and came down again so we could measure things like fuel temperature and heat. And then he went up and up to 50 feet and he held it. We looked at each other and said, ‘The Skunks did it again!’ and we hugged everybody. That was the real turning point. We knew the magic of the Skunk Works was still there.”

The lift fan’s success dashed Boeing’s hopes of an easy JSF victory. Yet the very same day at the testing base at Naval Air Station Patuxent River in Maryland, Boeing’s X-32A completed its first hover. As flight testing continued, with Boeing sometimes flying five missions a day and performing nearly flawlessly, the competition remained too close to call. Only a major slip would make one team the obvious winner.

At 1,500 feet over Patuxent River, Dennis O’Donoghue turned the X-32B downwind to prepare for its first vertical landing. As he brought Boeing’s STOVL demonstrator down to a stable hover at around 150 feet, he flicked a switch that turned on the jet screen—a narrow slot under the fuselage that blew cool engine bleed air toward the ground, helping to balance the airplane in hover and prevent the engine from sucking in its own exhaust.

By now brimming with confidence in the demonstrator’s handling qualities, O’Donoghue brought the X-32B gently down over the hover pit, a cavity in the runway designed to keep the exhaust from blowing back into the engine and minimize ground effect. The airplane coasted down to 40 feet, then O’Donoghue abruptly felt the bottom drop out underneath him. He jammed the throttle full forward and a red engine light flicked on, the automated voice he had hoped never to hear barking: “Warning, Warning, Engine, Engine.” Red lights were flashing in the control room too, and flight controller Howard Gofus tensely ordered O’Donoghue to abort the landing. But O’Donoghue was already at full power, and there was nothing more he could do. Bracing himself for a crash, he radioed, “I’m coming down!”

Then, with only seconds to spare, he got a reprieve. With around 20 feet to go, the engine recovered and the cockpit warnings ceased. The X-32B slowed its descent until it made a gentle touchdown at almost precisely its targeted landing speed. For the onlookers who rushed forward to congratulate O’Donoghue, it appeared to be a normal landing. Only Gofus and his team in the tower knew about the close shave, and they quickly figured out what the problem was. The wind that normally helped clear exhaust from the hover pit had momentarily died, and as the X-32B came down, its own exhaust gases had risen up from the walls of the pit and been sucked into the engine.

Confident in the diagnosis, Gofus went ahead with a second landing over a normal runway surface. O’Donoghue touched down without incident.

One week later, pilot Paul Stone of Britain’s Royal Navy took O’Donoghue’s place for another test: the X-32B’s first vertical takeoff. The flight plan called for a vertical landing followed immediately by a vertical takeoff. As Stone brought the airplane down gently, with the tires almost touching the runway, it wobbled momentarily, then a bright flash and a loud pop went off underneath the fuselage. Once again, hot exhaust had been sucked into the engine, and this time had caused a split-second engine “pop stall.”

The media were quick to seize on the episode, but O’Donoghue was unruffled. “All of us who worked with Harriers knew what that pop stall was and it was no big deal,” he says. “In fact, our simulations had predicted exactly what happened: If the plane tilted more than four degrees near the ground, the jet screen would no longer protect the engine and a stall would likely follow. We had already fixed the problem in our final design proposal with a bigger jet screen.”

As far as the Boeing team was concerned, the pop stall had been a nonevent, but the members knew it hadn’t looked good. Perception is everything, and the episode was a reminder of other inherent drawbacks that direct lift had and Lockheed’s fan didn’t. As with the Harrier, the 1,350-degree heat of the Boeing airplane’s exhaust gases would pose a threat to the surface of carrier decks, if not to the life and limb of Navy crews (the downdraft from Lockheed’s lift fan was some 1,000 degrees cooler). Since Lockheed’s fan boosted engine thrust, its powerplant could run at lower temperature and with less strain, and these differences would translate into longer life. Most significant, assuming its reliability could be ensured, the lift fan would offer an extra margin of power and safety in a hover. In the end, that ensured Lockheed’s victory.

The record of the flight tests offers an important clue to why Lockheed won. At the program’s conclusion, the X-35B had performed 38 STOVL flights, most at Edwards Air Force Base at an altitude of 2,500 feet. In contrast, Boeing’s X-32B had flown 57 STOVL flights, but these were all at the sea-level Patuxent River Naval Air Station. Despite the altitude advantage, the Boeing airplane flew with its inlet cowl and undercarriage doors removed to increase the thrust-to-weight margin. (“I would have left my underwear home that day,” says O’Donoghue.)

“In my mind, it was physics versus technology,” says Lockheed test pilot Paul Smith. “In the area of STOVL performance, Boeing just didn’t have the physics behind them—they didn’t have the thrust of the engine up and the weight of the airplane down, while we had a technology that made efficient use of engine power, but it was so technologically advanced that it was touch-and-go whether it would work. A month before we were supposed to demonstrate STOVL, we were still having problems with the lift fan that we thought we might not be able to fix. Boeing had done so many cool things, and were ahead of us on schedule so much. It was like the tortoise and the hare.”

On October 26, 2001, several hundred members of the Lockheed Martin team gathered in the X-35 hangar at Palmdale. On a big-screen TV, the Pentagon announced the company’s victory, causing the team to erupt in deafening cheers. “We did as much as we needed to win this thing,” an ecstatic Rezabeck told NOVA. “We were very comfortably—anxiously and nervously—confident!” In a contest in which both sides had displayed astonishing inventiveness, Lockheed had taken the bigger risk. And if the reliability of Bevilaqua’s lift fan has yet to be proven, it seemed reliable enough to win Lockheed the biggest military contract of the new century.

The X-35B lifts off the hover pit with its nozzle vectored for short-takeoffvertical-landing. To convert the engine’s operation from conventional takeoff to STOVL, the pilot moves a lever back about an inch. This opens four sets of doors behind the cockpit, allowing air to flow through the lift fan and starting the nozzle moving through its full range of travel. Simultaneously a clutch engages, transferring power from the engine to the lift fan. Heather Greasley/Lockheed Martin

Get the latest stories in your inbox every weekday.