by Tim Fernholz
Once in space, the Dragon folds out solar panels on either side, giving it a wingspan of forty-six feet. Its designers believe that the vehicle, when powered by the sun, can stay in orbit for as long as two years.
The flight was monitored in control rooms at Cape Canaveral and at SpaceX’s one-million-square-foot headquarters complex in Hawthorne, California, which had opened in 2008 to house the company’s growing manufacturing, design, and operations center. The young flight control team monitored telemetry feeds, watching for anomalous pressures and other problems as the Dragon edged nearer to the station. This was the first time a private spacecraft had approached the $150 billion global project in low earth orbit, and any accident could threaten not just a worldwide investment but the lives of the six men on board. For SpaceX, the hardest people to satisfy at NASA were the fiercely protective space station team.
After the Dragon entered the “keep out” area within a 650-foot boundary of the station, it became clear that there was a problem with the Dragon’s proximity sensors. These were a vital component as the vehicle attempted to approach the station at a safe distance and speed. The laser range finder that told the Dragon how far it was from the space station was giving a different answer than a redundant system that relied on heat measurements. Without both systems agreeing, the spacecraft couldn’t verify how far from the station it was—a serious problem.
This was not what SpaceX’s team had envisioned. A shakedown cruise eighteen months before this had established the capsule’s ability to traverse space, survive radiation and vacuum, and report back to earth. It had also carried a “secret payload”—an enormous wheel of cheese, another of Musk’s jokes. The company had planned on two more test flights: one to demonstrate the ability of Dragon’s communication link with the station, and a second flight to actually berth with the high-altitude laboratory. But, to save time and money, the Dragon team had proposed to NASA that if everything went well on the second flight, they could proceed with the objectives for the third mission right away. It was a page right out of George Low skipping intermediate tests to fly straight to the moon with Apollo 8.
“At first we didn’t receive it very well,” Mike Horkachuck, NASA’s official SpaceX minder, said of the plan to combine the tests into a single mission. “It seemed like they were looking for ways to just save money by eliminating a flight, and adding a lot more risk.” NASA eventually agreed to the accelerated schedule, in exchange for an expanded preflight test program that allayed some of their concerns about SpaceX’s system. The launch at Cape Canaveral had been interrupted while the company switched out an engine valve, but the Dragon had flown smoothly to its orbital perch near the station. After three days of successful maneuvering demonstrations and tests of the communications between the Dragon and the station, word came down from NASA: yes, it could approach the station and attempt a real rendezvous.
But now that a bug had been found, tensions were rising. Wouldn’t it have been smarter to test more in advance? Had SpaceX’s cost cutting put the station in danger?
This test flight was pivotal, and three years overdue. SpaceX hadn’t flown a single rocket since the last Dragon test, in 2010. The entire COTS program was now seven years old, almost twice the age of Gemini, and it had yet to deliver anything to the space station. “There was a lot of questioning, ‘Does this whole COTS thing work?’” Horkachuck said. “The political world is thinking, is this the smart thing to do? Should we really be messing with COTS and this whole commercialization, or should we be doing it the more traditional way?”
One SpaceX guidance controller acted quickly to buy more time, ordering a temporary retreat from the station. An objective of the communications test was to confirm that the astronauts would be able to press an emergency abort button within the station if something went wrong, which would fire the Dragon’s thrusters to send it safely away. If the astronauts had gotten nervous about the Dragon and aborted, another rendezvous wasn’t guaranteed. The retreat—“a brilliant maneuver,” in Shotwell’s words—bought time for the SpaceX engineers to solve the problem. They quickly realized that unexpected flashes of light from the station were interfering with the laser sensor. To fix it, they’d need to reprogram the software that interpreted the sensor data to ignore the extraneous input—and do it literally on the fly.
As the engineers gathered around computers to rework software, all the executives could do was wait. “Funny that it’s the term ‘berth’ with an e instead of an i,” Shotwell, who was in the control room, would later muse. “It was a little bit like giving birth, because it lasted a long time.”
Getting the Dragon to that point—a few hundred feet away from the ISS, tantalizingly close to fulfilling the mission it was assigned six years before—had replaced the Falcon 1 as SpaceX’s main challenge. Indeed, the rocket had been the easy part: the Falcon 9 rocket that launched the Dragon had flown for the first time just months after President Obama paid his respects on the launchpad at Cape Canaveral. Despite the anxiety stemming from the company’s explosive efforts in the Pacific and the reality that most new space vehicles fail early, the launch had gone swimmingly. The Falcon 1 experience had prepared SpaceX’s team for their next challenge: teaching them about their technology and how to use it.
Some outside observers fretted that flying with nine engines violated the rule, adhered to by Musk, that a simpler system is always a more reliable one. But in this case, Musk and the other designers saw it as redundancy: if one engine failed, eight others would suffice to keep the rocket flying. It was also far cheaper than building a new, huge engine, which had been the traditional approach American rocket builders took to the problem.
“Engine cost is not linear with thrust or performance level,” Shotwell explained to me one afternoon in SpaceX’s DC offices, meaning that a bigger engine is not necessarily cheaper or more powerful. “If you have to build a giant engine, that’s going to be way more expensive than building nine smaller engines. You build an engine that’s the size of this room, think of the machine that you would have to build it in. Think about: how do you fire that engine? You’ve got to have this giant hoist to get it into the test stand. Think about buying a billet of aluminum . . . How do you produce that giant block of metal that’s capable of being the constituent part of an engine? It’s really hard.”
Using nine Merlin engines, plus another to power the second stage, had another bonus: it meant lots of engines in production, generating economies of scale and reliability. The company leveraged this across the board, using the same tools to make the first and second stages of the rocket. They adopted 3-D printing for critical components sooner than their competitors. Between the advantages of SpaceX’s low-mass rocket and high-powered engines and its approach to doing business, the Falcon 9 would be the cheapest orbital rocket on the market—if it could fly.
Though the company had progressed rapidly through its initial design milestones for the COTS program, as it began manufacturing and integrating actual hardware, it ran into new problems to fix, as well as associated delays. The first Falcon 9 launch, while successful, was eighteen months behind schedule, and that meant the company hadn’t received the associated milestone payment. Financially speaking, this was fine for NASA, but it meant that SpaceX was paying out of pocket for its own delays.
There was even a small measure of revenge from the canceled Constellation program: SpaceX had planned to use the same parachutes as NASA’s Orion spacecraft, but by the time the Dragon was ready for them, they had not been certified for human spaceflight. To leapfrog NASA, SpaceX needed to create its own testing program to ensure that the parachutes worked, and dropped a Dragon from a helicopter fourteen thousand feet above the Pacific Ocean.
Now that SpaceX was getting paid more than $1 billion by the government, the scrutiny increased. Like Griffin and the Constellation program, SpaceX was subject to the searching inquiries of the Government Accountability Office, which had warned in 2011 that neither they nor Orbita
l would be likely to meet the deadline for flying cargo to the space station. This forced NASA to extend its expensive agreements with its international partners to fly supplies and astronauts to the orbital lab.
Musk worked to protect his company’s culture, warning the space agency that “for every NASA person you put on my site, I’m going to double the price.” There were inevitable clashes, some as simple as NASA’s love of acronyms, which Musk loathed as an obstacle to understanding, unless deployed with crude irony.
“I read the NASA document, and it’s so full of acronyms I can’t understand it. I go, ‘I have no idea what that means,’ and I’m working in the same area,” chief launch engineer Koenigsmann lamented. There were pitched battles over documentation, which for SpaceX meant dynamic electronic records, and for NASA copious redundant printouts. Sometimes it was simply about getting counterparts on the phone. “For other contractors or government organizations, you have to consider that there’s no way to call them after Friday at 3:00 p.m.,” Koenigsmann said. “There’s a lot going on here on Friday at 7:00 p.m. at SpaceX.”
Despite these communication issues, there was still a productive back-and-forth between SpaceX and the space agency, with the government nudging the company toward more planning and documentation. “I spent a long time trying to get them to actually build a schedule, because they had really no idea how to lay out a big project schedule when we started,” Horkachuck remembered. As SpaceX evolved its designs and worked to combine the last two test missions, NASA (armed with the seal of approval from the Augustine Commission) convinced Congress to deliver an additional $300 million to the COTS program for further safety testing.
“As we added all those more traditional tests that NASA would have done on a system, we got more comfortable with them being able to combine the two missions and not being so risky that it was just a throwaway,” Horkachuck said. “Some of those big system tests found real problems that would have been a mission failure if we didn’t catch them on the ground.”
With the money, SpaceX stuck the Dragon into vacuum heat chambers to make sure it held up to the extreme environment of space, and also subjected its computer systems to electromagnetic interference to make sure that they wouldn’t short out in orbit. The company even flew its laser range-finding sensors on one of the last space shuttle missions so that its hardware could get a chance to operate in space before it became critically important to succeed.
Still, SpaceX brought its own approach to designing the capsule. Instead of working through computer simulations to test the ability of astronauts to move around inside the capsule and unload it, the company built a mock-up and found two employees who were of appropriately average size to scramble around inside. (This saved the engineers from doing a casting call for out-of-work actors to test the system.) Rather than use light meters to ensure that labels were bright enough to read, they simply had an astronaut come and look at them. They saved $1,470 on each locker handle by using bathroom-stall latches, and chose NASCAR racing safety belts over custom-built aerospace harnesses. It turned out the NASCAR belts were just more comfortable, perhaps because the drivers spend more time belted in their seats than astronauts do. The Dragon launched with a nose cone to improve its aerodynamics and protect the docking mechanism. To test, as cheaply as possible, the nose cone’s ability to pop off after reaching orbit, the Dragon team bought a children’s bouncy castle and ejected the cone inside. It worked just fine.
There were also fruitful collaborations between SpaceX and the space agency. While the Constellation program couldn’t figure out how to replicate the heat shield from the Apollo mission, SpaceX worked with NASA’s Ames Research Center to adapt a more recently invented material called Pica. It had been created to enable scientists to return a sample taken from a comet, a mission requiring the space probe to survive a return through earth’s atmosphere while flying 29,000 miles an hour.
Through the COTS program, SpaceX was able to bring Dan Rasky, one of the material’s inventors, temporarily in-house to make the thermal shielding cheaper and more resilient. The resulting material, known as PICA-X, was a real innovation. During the Dragon’s first superheated plunge through the atmosphere, the heat shield worked so well that there was concern that some sensors might be broken—they showed no change in temperature throughout the entire descent.
But all this work took time, and the pressure was building. As the first flight to the space station approached, SpaceX and NASA teams were pulling all-nighters to ensure that everything would go smoothly. The aberrant proximity sensors were no small issue: SpaceX used a more modern approach to software engineering that relied on constant iteration, while NASA wanted a review of every single change; before the mission, the two organizations had gone over more than twelve hundred individual changes to the flight code. Now, once they finished their reworking, they would need to run the final product through NASA simulators before they could upload it to the vehicle in orbit.
“You couldn’t possibly model it. Use your best guess,” Lindenmoyer told me later, describing the engineers taking the imagery, manually filtering out the bad data, and figuring out how to get the two sensors to converge. “It was a beautiful piece of engineering work, under stress, and just showed the tenacity and the agility of this company to get it done.”
The code passed NASA’s tests, and in the wee hours of the morning it was uploaded to the Dragon. It resumed its approach to the station, this time with both proximity sensors on the same page. Once it was just thirty feet away, astronaut Pettit made good on his training by swiftly grabbing the vehicle and pulling it tightly to the station’s airlock. “Looks like I have a Dragon by the tail!” he told NASA Mission Control in Houston.
“If I gooned it up, it could set commercial spaceflight back for years,” he told me later. “The bottom line is, the engineers at SpaceX did an amazing job of designing the control system for Dragon. It was a cream puff.”
When the astronauts popped open the hatch, they reported that the Dragon smelled just like a “new car.” It was the first privately developed spacecraft to dock with the International Space Station, or indeed any space station.
Outside the control room, staffers who had stayed late to witness the berthing started screaming, crying, and hugging. Many had been working in the control room since the previous day. There were tables full of champagne and flutes, and Shotwell started popping bottles for her team as the company’s celebrations began. These hobbyists had now not just gotten to space, but proven that they could do something lucrative there. The credibility gained by SpaceX’s successful integration into NASA’s stringent approach to human spaceflight was immeasurable. What’s more, they had taken NASA back to the space station—the first US spacecraft to visit the lab since the final shuttle flight the previous year.
“Quite frankly, if that flight was not successful, I’m not sure a commercial crew would have taken off,” Horkachuck said afterward, referring to the follow-on program that would allow private companies to fly astronauts as well as cargo.
The COTS program had paid SpaceX $396 million to develop the Dragon spacecraft and Falcon 9. SpaceX says that it contributed $850 million to the development effort. During that period, the company raised just $220 million from its backers. That left some $630 million that was generated by follow-on contracts from NASA, but also the endless road show led by Shotwell that generated a lucrative launch manifest with dozens of missions (and accompanying deposits). In 2014, the company reported a manifest of thirty-seven future launches in one legal filing. However it was financed, the production of new rocket engines, a new spacecraft, and an orbital rocket for $1.2 billion was a feat of aerospace business.
Before releasing the Dragon to return to earth, Pettit and the other astronauts took a self-portrait inside the capsule, made a print on the space station’s aging and perpetually low-on-ink printer, and autographed it. Pettit had one of his colleagues float up to block a camera monitoring their activities so he
could sneakily tape the photo into the capsule as a surprise to SpaceX’s recovery team.
“They have that crappy printed picture where all the crew had signed it properly framed and hanging up in their hallway,” he told me.
11
Capture the Flag
Going through test pilot school, there isn’t a student who doesn’t think the dream job would be to be a flight-test engineer on a brand-new spaceship and then get a chance to go fly on it.
—Astronaut Robert Behnken
Blue Origin’s first real step out into the public eye came in 2010, thanks to the Obama administration’s enthusiasm for commercial space exploration. To kick-start the next stage of its commercial partnerships, this time focused on flying astronauts to the space station, NASA put up a small pot of money for a program called Commercial Crew Development, or CCDEV. (NASA loves acronyms.) The first $50 million was part of the nearly $100 billion stimulus legislation the new president devised to goose the flagging economy, and NASA put a share of that money into the commercial program.
The first round of funding went to companies working on the technologies needed to bring humans up to the ISS. Boeing and United Launch Alliance received funds to upgrade their spacecraft and rocket designs, respectively, to fly humans safely. Sierra Nevada Corporation (SNC), a satellite-and-propulsion-technology company, received $20 million. They were developing a shuttlelike spacecraft called Dream Chaser, a former NASA design that had been upgraded by another space company before falling into SNC’s hands. Blue Origin received the second-smallest grant of all, $3.7 million, to work on the New Shepard. The money went toward developing an airtight carbon-fiber capsule that would transport the participants, and a “pusher” launch escape system, which would allow a capsule full of tourists to jet away from harm in the event of danger from the booster rocket below.