SpaceX takes a fresh approach to rocket fabrication
November 11, 2008
Managers, designers, and manufacturing engineers at Space Exploration Technologies have come up with a new way to design and manufacture a rocket.
This is a rendering of the Falcon 9 rocket SpaceX plans to launch next year. Image courtesy of SpaceX.
Aerospace executive Chris Thompson, who reports directly to his company's CEO, knows all about the retractable-pin friction stir welding system on the floor. He directly oversees it and has operated it himself.
That's not typical for a vice president in the aerospace business, but Thompson doesn't work for a typical aerospace company, or a typical boss. He's vice president of structures for SpaceX, a company that earlier this year moved into a half-million-square-foot Hawthorne, Calif., facility near L.A., the same building where Vought Aircraft Industries once fabricated Boeing 747 barrel sections.
Thompson reports to Elon Musk, the Internet tycoon who sold PayPal to eBay for $1.5 billion in 2002. Instead of pocketing the earnings and retiring comfortably, Musk used it to get back into the entrepreneurial trenches. Earlier this decade, barely into his 30s, Musk founded Space Exploration Technologies (SpaceX for short) with the ambitious goal of launching satellites for about a quarter the price offered by the world's government space agencies. He planned to do this by marrying what he's learned from his Internet experience with manufacturing, engineering, and design talent he hired away from legacy aerospace players.
Together, they've come up with a new way to design and manufacture a rocket (see Figure 1).
The new way is apparently working. When on Sept. 28 Falcon 1, SpaceX's first and smallest rocket, launched successfully into orbit from the Marshall Islands' Kwajalein Atoll in the middle of the Pacific, managers said it marked a major milestone. Falcon 1 can carry a 1,400-pound payload, enough for a small satellite. With a price tag of less than $10 million (inexpensive by rocket standards), the vehicle may open up the market to various organizations and people, including university researchers, who couldn't afford sending a payload into orbit before.
To attain orbit at such a low price required some unorthodox approaches, sources said. Rocket designers haven't borrowed from current designs, but instead are starting from scratch, hoping to simplify one of the most complex human endeavors.
Thompson described a conventional rocket design of machined plates formed into a barrel section. It saves weight, but it's costly and time-consuming to manufacture. As an alternative, "you can do a really simple structure: a welded tank of rolled aluminum—super-simple, super-inexpensive. There is some mass inefficiency there, but your labor inputs are much lower, and overall you draw the costs down significantly."
The aluminum-lithium tank of the Falcon 9 rocket sits in a cradle at the SpaceX Hawthorne facility.Photo courtesy of SpaceX.
SpaceX has a corporate structure that, according to sources, supports collaboration and efficient decision-making. A designer with an idea can walk over to the manufacturing engineer, talk about it, and then go to the floor to see if it will work.
Thompson added that this couldn't happen without another unorthodox strategy: in-house manufacturing. The 500-plus-employee firm doesn't require high volumes; in its early years SpaceX found outsourced jobs delayed as job shops gave priority to larger contracts. So the company, mostly an assembler in 2002, since has brought 90 percent of its manufacturing, including almost all of its metal fabrication, in-house.
On SpaceX's vast floor, workers perform bump forming on press brakes. Tank sections are rolled on a massive four-pinch rolling system with an 18-in. throat. There are the typical arc welding processes along with some less common joining methods, including friction stir welding of some exotic aluminum and aluminum-lithium alloys for rocket tanks (see Figure 2). The circumferential friction stir welds are done using a retractable-pin head from Nova-Tech—the same kind of weld head used for the Shuttle program—while SpaceX engineers designed the fixturing and drive mechanisms in-house. The longitudinal welding system, from Transformation Technologies Inc. (TTI), doesn't have a retractable pin and so welds run into a sacrificial tab. "We weld anywhere from 0.063 inch up to more than half-inch material," Thompson explained.
For engine production the company has several electric-hydraulic tube benders, including one that bends up to 3-in. INCONEL® alloy tubing. And there's an extensive machine shop to cover the milling, turning, electrical discharge machining, and some arc welding work.
Manufacturing responsibility is organized so that engineers and manufacturers can work together. September's successful launch was a fourth attempt, something that points out the obvious: Space travel isn't easy. So collaboration is important, sources said, because problem-solving must happen efficiently.
Thompson recalled one Falcon launch in which there was a separation failure between stages. Engineers traced the problem to corrosion on a nut. "So we ran through the basic fault tree to understand the problem. Can we switch that to an orbital tube weld and get rid of that field joint? Is there a different way of securing it? If we have to make physical changes to the hardware, it's done quickly and tested carefully. We don't get into an infinite loop where you analyze things nine ways to Sunday."
The two chiefs over the lion's share of metal fabrication are Thompson, vice president of structures, and fellow engineer Robyn Ringuette, director of propulsion production, who oversees the company's demanding tube bending operation, among other things. Much of the rest of manufacturing falls under the company's machine shop, which reports to its own VP.
The organization marries engineering with critical manufacturing functions. Those working with tube bending, orbital welding, and friction stir welding are grouped directly under the engineering group that designs parts for them. Thompson oversees friction stir welding directly because the process intimately affects the quality of the final structural product, including the Falcon's aluminum-lithium tank walls.
Ringuette oversees tube bending and orbital welding because these processes relate so closely to how the propulsion system works. Orbital welding closely ties in with the bending operation, too, because issues with one—be it production rate, tolerance, or setup problems—can directly affect the other.
"The guys out there bending tubing are intimately involved with the process," Ringuette explained. "It's not just 'fed down' from engineering."
Thompson came from McDonnell Douglas (now Boeing), and Ringuette came from Rocketdyne Propulsion & Power, now owned by Pratt & Whitney. Neither said they've reported to a boss quite like Elon Musk.
"I meet with Elon regularly," Ringuette said. "He's an intense person."
But he does promote an atmosphere of collaboration. A video tour of the engineering facility on SpaceX's Web site, for instance, shows Musk, in a polo shirt and jeans, walking through the company offices, only there are no offices—just an expanse of low-walled cubicles.
"That's my office over there," he said, pointing to a cube area in the corner. "We try to minimize the number of offices we have. Doors limit communication. Everyone at the company, with the exception of those in HR and finance, are in cubes, including the vice presidents."
The VPs include Thompson and Ringuette, and both said they appreciate the lack of bureaucracy.
As Ringuette explained, "When I need to buy a new machine, I describe it to Elon, he either agrees or disagrees, and that's the end of it. It's very helpful, because it keeps me focused on finding the right tools for the job. Nothing is more complicated than it needs to be."
This included Ringuette's hunt for a tube bender.
A Merlin engine from SpaceX undergoes a test. Photo courtesy of SpaceX.
Ringuette is responsible for producing what sends the Falcon rockets into space: the Merlin engines (see Figure 3).
"Anything that goes into propulsion that's engineered at SpaceX, it's my problem," he said.
Each engine requires a tremendous amount of tube. When bending was brought in-house several years ago, tubes were actually hand-bent and hand-flared. "That gets old pretty fast," Ringuette said, "so two years ago we brought in some Pines Technology machines to bend 1/4-inch to 1-inch stainless steel, titanium, and some aluminum. We just brought on an Eaton Leonard machine that can bend from 3/4 inch to 3 inch."
This year the company upgraded its bending capability to CNC, so "the tubing is completely modeled and defined in 3-D space," Ringuette said. "The bend table that defines the coordinates is output to an electronic file, and that's input into a machine.
"Everything is linked with CAD/ CAM," Ringuette continued. "We still have drawings to verify the file is correct. But raw data is not being punched in by the operator. That eliminates a lot of errors at a machine—an operator fat-fingering a 5 instead of a 2, for instance."
After bending, a laser scanner checks to see if the bend matches the print. From there the workpiece is cleaned and then joined with orbital welding units from Arc Machines Inc. and others. It then goes into hydrostatic pressure testing and helium leak testing before being sent on to engine assembly.
The company has some hefty bending requirements, working with material like 316 stainless, titanium, and, the most challenging, INCONEL. Certain lines must endure gas that's heated to 400 degrees F and pressurized to 5,000 PSI. "It gets pretty toasty in there, so we need some thick-walled stuff," he said.
This bending head on the Eaton Leonard VB 80 HP tube bending system is similar to the one used at SpaceX. Photo courtesy of Eaton Leonard.
SpaceX's Eaton Leonard machine is put through some stringent demands, bending 3-in. INCONEL tube to a 3-in. radius—the dreaded 1xD, where the radius matches the workpiece diameter. As Ringuette put it, "In our industry, people like to see 3xD, and they grumble a bit if you're at two times diameter, and here we are bending 3-inch tubing on a 3-inch radius. It's a lot of stress on the machine, and it's a lot of stress on the tool."
As Bob Wolbrink, regional sales and marketing manager for Eaton Leonard, explained, "Such bends require so much force on the tube as the machine draws it around the radius. It's going to want to slip, so you need to have adequate clamp force against the tube."
The unit, a VB 80 HP, has an electric motor for the bend drive and a hydraulic clamp (see Figure 4). Various elements make that large-diameter INCONEL bend possible. First is fixturing and fit-up. As Wolbrink explained, "You need a long enough grip" to hold the tube and ensure it doesn't slip. Also, tight clearance between the mandrel and tube ID is paramount. As Wolbrink explained, "Any excess of clearance gives room for the material to deform and wrinkle onto itself. And the wiper die [a kind of wedge that goes between the bend die and the tube] must be machined properly to give the right amount of support." The proper application and amount of lubrication helps as well.
Although Ringuette's meeting plugging the need for a bigger tube bender went quickly, Thompson's push to bring in friction stir welding didn't. But today the benefits of having FSW systems on the floor are paying off.
"That was a tough sell," Thompson recalled. "The process has a very steep learning curve, but the benefit in the end is that 99 percent of the welds you're producing are defect-free. We found also that friction stir welds are relatively easy to repair, compared to the conventional grinding, preparation, and rewelding. With friction stir welding, you just restir it."
Conventional welding required an X-ray inspection that forced Thompson to shut down the welding area. The new FSW system ultrasonically inspects the welds concurrently with the weld process, following several feet behind the welding head. The welding also undergoes die-penetrant tests postprocess, but by the time the inspectors move in, the FSW system is already on to the next weld.
A two-engine test for the Falcon 9 lights up the evening sky. Photo courtesy of SpaceX.
Future milestones include the launch of SpaceX's next generation of vehicles, the Falcon 9, with the maiden flight planned for 2009, and the launch of the even bigger Falcon 9 Heavy in several years. The company conducted engine tests for the Falcon 9 in August (see Figure 5). Beyond this, SpaceX has bigger plans that include Dragon, a capsule that eventually could send cargo and even passengers into orbit to dock with the International Space Station once the Space Shuttle retires in 2010.
"Falcon 9 is the most powerful single-core vehicle in the American fleet," Musk said in an August Webcast on the company Web site. "When Falcon 9 Heavy debuts in about two to three years, it will be the most powerful vehicle in the world, because the shuttle will have retired by that point."
The Falcon 9 will be able to haul 22,000 pounds. The market for such payloads is huge, sources said, and to meet demand SpaceX plans to launch a Falcon 9 a month.
That's right—a rocket a month. And the company seems well on its way.
"This year SpaceX will make more rocket engines than the rest of U.S. production combined," Musk said during the same Webcast. "Next year we'll make more rocket engines than any country in the world."
Engineers are transforming SpaceX's manufacturing operation from an experimental prototype shop to a production facility, and the company's significant capital equipment acquisitions are part of the plan.
"With the launch rate we advertise, it's 10 times that for engine production [on the Falcon 9]," Ringuette said, explaining the Falcon 9 uses nine engines on the first stage and one engine for the second. Add to that the ancillary plumbing that goes with those stages, as well as the lines necessary for the Dragon capsule (which the Falcon 9 will be able to carry), and "you've got a lot of tubing," he said.
Despite the daunting challenges, employees at SpaceX, like their boss, still tend to think big. September's launch represented the first liquid-fueled rocket to be put into orbit by a private company—not a small accomplishment by any measure. But Musk and his team have bigger plans on the horizon.
"In the coming years, we see us as being the leaders in the world launch market," Thompson said.
He paused, and then added, "This is really a dream job. You're not dealing with legacy programs. Instead, you're starting from a clean sheet of paper, and you're in charge. And you get a chance to be a part of history."