by Tim Fernholz
“You’re going to have to land a two-story house on Mars if you’re going to send humans,” Bobby Braun, a former NASA chief technologist who is now dean of the University of Colorado’s engineering school, told me regarding landing schemes to survive on the Red Planet. Furthermore, he said, a Martian colonizer must land “right next to another two-story house that’s been prepositioned and powered up and has all the fuel and food that humans will need to survive on Mars.”
Putting heavy objects on distant planets has proven extraordinarily difficult for NASA. The heaviest mass sent so far is the nearly two-thousand-pound Mars Curiosity rover, which landed in 2012. To reach the surface, it required a Rube Goldberg contraption that included heat shields, parachutes, and a rocket-fired crane that finally lowered the rover to the ground. The space agency’s engineers had been afraid to use rockets earlier in the landing process—at high altitude, flying at supersonic speeds—because they didn’t know enough about how such a vehicle would perform. When SpaceX began flying the Falcon 9 at supersonic speeds in earth’s atmosphere, the company shared its data with NASA scientists who were plotting missions to Mars. They were eager to receive it.
“With supersonic retropropulsion, there was no reason to believe it would not work. But there was no reason to believe that it would work,” Miguel San Martín, one of those researchers, told me. “In the culture of NASA, we were going to do a big testing program. Elon Musk just tried it. And if it works, it works.”
In 2011, after the company had flown its Falcon 9, SpaceX hired an engineer named Lars Blackmore from the Jet Propulsion Lab. A product of MIT, Blackmore was an expert in designing software for autonomous vehicles to navigate extreme environments; one academic project had guided a deep-sea submersible robot, and at JPL he wrote a critical algorithm to guide landers arriving on Mars. His graduate adviser, a NASA veteran himself, said Blackmore would have made a tremendous professor, but he went to SpaceX because it offered “an opportunity for the current generation of engineers to make their vision real.” At SpaceX, his job was to teach the Falcon 9 to come back to earth in one piece.
That year, Blackmore began work at the Texas launch range on the SpaceX project called Grasshopper. It, too, was reminiscent of the DC-X, involving a small prototype of a rocket that could launch, hover, and return to earth. The Grasshopper was just a hundred feet tall and mounted on metal struts. In September 2012, it took its first hop into the air; a year later, it flew three-quarters of a mile in its final test. By 2014, the reusability engineers were using a 130-foot-tall, full-scale first stage of the Falcon 9, equipped with four space-rated, retractable landing legs. They sent it as high as 3,300 feet in the air to hover before returning to settle gently down on the landing pad. During one test, the landing legs actually caught fire, delivering a biblical image of a flaming sword in the air for any passing eschatologists. During another test, a blocked sensor caused the rocket to veer away from the safe area; automated software blew the rocket up to prevent it from endangering anyone. The explosion attracted local interest and press criticism, but it did not daunt the engineers.
These experiments taught them—and the algorithms controlling the rocket—valuable lessons about how to adjust the engine to compensate for changes in the vehicle’s position and the surrounding conditions. They adapted complex mathematical software developed by Stanford computer scientists so that the guidance computers could find a safe path back to earth with only a tiny margin of error—small enough to guarantee a soft landing inside a sixty-five-foot ellipse on the ground below.
Still, the calm conditions and slower speeds of the test range were very different from what a rocket plunging toward earth from space endures. They needed more data, and the pragmatic SpaceX team had already been gathering it during operational missions. After the Falcon 9’s first satellite launch mission, in 2013, the booster stage had a secondary mission of its own: steering back down to the ocean. While it didn’t make it all the way—it lost control and smashed into the Pacific—the reusability team garnered valuable information about how to fly the unusually shaped vehicle.
The open ocean wasn’t just a safe place to crash a rocket tumbling back down from space. It was also where the rocket would have to land in order to make reusability work. Strenuous calculations had confirmed that, while it might be ideal to return the booster rocket to earth in the vicinity of its launchpad, the physics of doing so would prove prohibitive. Rockets do not go straight up when they take off, but rather turn to fly across the earth and into their chosen orbit at high speed. For missions to destinations in low earth orbit, like the International Space Station, the rocket would have enough fuel left over to fly back to its landing point. But for missions further up into space—and these were more numerous and more lucrative—the rocket would need to use almost all of its fuel simply to get the job done. After that, unable to return to land, the rocket could be saved only if it returned to earth offshore—say, for example, on a floating landing pad.
This idea was why, in 2014, SpaceX challenged Blue Origin’s patents in court. In general, SpaceX did not believe patents would be useful for protecting its intellectual property; Musk saw them mostly as a way of telling competitors—especially those outside the United States—exactly what he had done that was so unique. Blue, on the other hand, seemed to love patents.
One of the public signs of Blue’s reemergence after 2010 was a proliferation of patent filings on exactly the kind of components needed for reusable rockets—steerable engines, methods for lightweight construction, and guidance techniques. Just as lifting the heavy equipment needed to colonize Mars motivated SpaceX’s desire for reusable rockets, it was equally important to Bezos’s goal of shifting industrial capacity into orbit, followed by human civilization writ large. Bezos holds numerous patents related to Amazon’s marketplace and subscription services, but he has put his name on only one of Blue’s: “Sea landing of space launch vehicles and associated systems and methods.”
The patent is for a reusable space vehicle taking off over the ocean, launching its cargo, then turning its engine back on to descend onto a floating platform. This was exactly what Musk and his team intended to do with the Falcon 9, and SpaceX’s attorneys became concerned that even if they beat their competitor to the punch—as seemed likely—they would be vulnerable to litigation. To preempt this situation, they challenged the patent in court, and in doing so offered a brief lesson in the history of the idea to show that it hadn’t originated with Blue, or SpaceX, at all. It had been described in some detail as early as 1998, by a Japanese engineer, Yoshiyuki Ishijima.
It was another clash between the two rocket billionaires, and again it was Musk who came out on top. In early 2015, the judges who reviewed SpaceX’s challenge found that the bulk of Blue Origin’s claims were too broad to be suitable for a patent. The judges declined to review the remaining two provisions, because the descriptions were too “indefinite” for them to determine whether SpaceX would prevail. While this entailed a rejection of SpaceX’s request, it was, in effect, a victory: a patent deemed “indefinite” would be vulnerable to challenge in federal court, and the ruling provided a measure of protection for SpaceX against future litigation by Blue. Now it was simply a matter of actually landing the rockets.
In a series of launches in 2014, SpaceX honed the reusability of the Falcon 9. Its engineers repeatedly guided the rocket down to hover above a specific spot in the ocean and then deploy its four landing legs, before it ran out of fuel and sank into the waves. In early 2015, the company unveiled two new pieces of technology. One was grid fins—four metal honeycombs, about five feet square, mounted on the sides of the rocket. Originally used on ICBMs, they have the ability to rotate and maneuver the rocket by altering the flow of air around it.
The other debut was two autonomous drone ships—large barges capable of operating without human crew so that they could safely function as floating landing pads for the rocket booster. The company had one for each ocean�
�at Vandenberg, on the Pacific, was a ship called Just Read the Instructions, and at Cape Canaveral, on the Atlantic, was its partner, Of Course I Still Love You. The two names came from a science fiction series, favored by Musk, about hyperintelligent AI spacecraft plying the stars.
On its fifth mission to the ISS, which went off without a hitch, SpaceX attempted to land a rocket on Of Course I Still Love You for the first time. Musk had warned the press that this was just an experiment and the company didn’t expect it to succeed—and, boy, did it not. The hydraulic system controlling the grid fins ran out of fluid due to the number of course adjustments required. The booster came in sideways at high speed, caroming off the edge of the floating platform and bursting apart over the waves. After another flight, in April, the rocket touched down on the floating platform almost gently, but a sticky engine valve shut off the engines just a bit too late—the extra momentum made the rocket tip over and blow apart, scattering debris into the blue water. After a later near miss, Musk would adopt a euphemism for its consequences: RUD, or “Rapid Unscheduled Disassembly.”
SpaceX shared videos of these misses on social media; the varied explosive attempts to land the Falcon 9 first stage on the drone ships would become iconic among SpaceX fans and employees, populating a blooper reel of rocket test disasters. The company’s decision to stream its launches live on the internet was unusual, as was the decision to tap the company’s actual engineers to explain each step of the launch in some detail. As a public relations strategy, it helped underscore how far the company was pushing the limits, even if it created opportunities for people who didn’t understand that these were experiments—or who deliberately ignored that fact—to call them out as failures.
Blue Origin took the opposite approach to publicity, conducting test flights in secret and only then announcing the results. In April 2015, weeks after SpaceX’s booster tipped over at sea, Bezos’s team finally had some results worth sharing. The company reported that it had launched its New Shepard vehicle, which had been mooted since 2003, for the first time. With Bezos in the control room, the company elevated the stubby rocket and capsule, with Blue’s feather logo painted across the entire booster, to its full height of fifty feet. The rocket’s engine ignited and it flew fifty-eight miles into the air before the empty capsule separated, soaring on a ballistic trajectory to the edge of space before it fell back to earth, deploying three parachutes to land safely in the desert.
Bezos, in an update published on the company’s website, said the test would have been flawless—if they had been building an expendable rocket. “We didn’t get to recover the propulsion module because we lost pressure in our hydraulic system on descent,” which is another way of saying that they dropped the New Shepard onto the ground. It was similar to the fault that betrayed the Falcon 9’s first sea-landing attempt—a sign that both entrepreneurs were zeroing in on the tools they desired as the terms of their unspoken race became clear.
Yet the two companies were offering a very big difference in scale. The New Shepard was a marvel of engineering, to be sure, but it was about as powerful as the Falcon 1 had been seven years earlier, and without a second stage to allow it to accelerate satellites to orbital velocity. New Shepard reached three times the speed of sound in flight; the Falcon 9 flies six times faster than sound or more. “True spaceflight is when you need a rocket to get you back down again,” space historian David Woods observed in one interview. Blue Origin’s vehicle didn’t experience the enormous forces faced by the much larger SpaceX vehicle blasting through the atmosphere with a million pounds of force behind it, and it was a commensurately smaller achievement. Perhaps because of this, Bezos noted that his team would be applying the lessons learned on the New Shepard to a much larger rocket, which would be powered by the engine Blue was building for United Launch Alliance’s next rocket.
At the same time, though, the difference in scope between SpaceX and Blue Origin was intentional, and telling. Bezos’s company was not bootstrapping itself into the future with the markets that already existed in space, as Musk’s team had with satellite launch. Thanks to its founder’s enormous wealth, Blue could aim to create a new market for space tourism that had never existed before. In 2015, the company started signing up interested parties to a mailing list that was advertising its “astronaut experience.” The New Shepard was perfectly designed to deliver it: an almost gentle rocket that could introduce the public to space in a capsule that, the company promised, had fully a third of its surface covered in windows—an idea that Blue’s executives trace to the same “relentless customer focus” that Bezos calls for at Amazon. Now he could win a new market, but only if his rocket could be reused successfully enough to drive down ticket prices, and safely enough to convince people to get on board. And there were plenty of warnings about hubris in commercial spaceflight.
14
Pushing the Envelope
The first time I took a week off, the Orbital Sciences rocket exploded and Richard Branson’s rocket exploded . . . The second time I took a week off, my rocket exploded. The lesson here is don’t take a week off.
—Elon Musk
When the Falcon 9 exploded on June 28, 2015, on its seventh mission to the International Space Station, an instant wave of queasiness cascaded all the way from the control room in Cape Canaveral to SpaceX headquarters, in Hawthorne. Thousands of fans were watching the company’s livestream of the launch on YouTube and saw the rocket break apart in the atmosphere, just over a minute and a half into the mission. It was the nineteenth flight of a Falcon 9 rocket. It was also the first real operational failure in SpaceX’s history—each previous screwup had in some way been an experiment.
“The hardest but best approach is that you pick up the phone and you call them right away,” Gwynne Shotwell would recall. “You can’t avoid that. You blew up the rocket and you blew up the Dragon.”
And perhaps the hardest problem is that you don’t have anything else to tell them. The high-speed nature of rocketry and the complexity of the machine make it impossible to know what has gone wrong until engineers have spent several weeks going over data feeds—three thousand of them in this case—with a fine-tooth comb. They watch video footage of the launch from multiple angles, including from cameras that NASA installed to monitor launches in the wake of the Columbia disaster. And they examine any wreckage they can recover from the ocean.
After the failure, as the US Air Force and Coast Guard secured the range and the ocean underneath the explosion, emails flew back and forth between Musk and his engineers, attempting to drill down into what went wrong. Already, SpaceX’s team was collecting information and testing theories.
“It’s terrible, but it’s not unheard of, that rockets blow up—you’re taking a million pounds of explosive force and trying to shove it down that way—you don’t want it to go that way!” Shotwell told me with a sideways swipe of her hand. “In order to have that million pounds, you’ve got high-pressure helium systems which shove the propellants into the engine—it’s just hard.”
An hour and a half after the accident, Musk tweeted, “There was an overpressure event in the upper stage liquid oxygen tank. Data suggests counterintuitive cause. That’s all we can say with confidence right now. Will have more to say following a thorough fault tree analysis.” And because Elon is Elon, he also found time to reply to a fan’s condolences with a brief “Thanks : ).”
The explosion was a major problem not just for SpaceX and NASA but for the broader idea of “space as a service.” It was the second strike against the space taxi program. The previous fall, in October 2014, Orbital Sciences had flown the Antares rocket and the Cygnus space capsule, which it had built with NASA’s development funding, for the fifth time. During a launch at Virginia’s Wallops Flight Facility, a turbopump in the engine cracked just six seconds after liftoff. The resulting conflagration blasted apart the vehicle as well as the launchpad.
Like Lockheed Martin and the Atlas V, Orbital had also gone the
route of surplus Russian engines, using the NK-33. These engines had been state-of-the-art—when they were designed, back in the 1960s. Since then, dozens had been stored away in warehouses. Several different US rocket design programs had considered using them before Orbital selected them to drive Antares.
This didn’t impress Musk, who, at ten years into his self-education as a rocket engineer, was confident enough to mock the decision. “One of our competitors, Orbital Sciences, has a contract to resupply the International Space Station, and their rocket honestly sounds like the punch line to a joke,” he told Wired magazine. “It uses Russian rocket engines that were made in the ’60s. I don’t mean their design is from the ’60s—I mean they start with engines that were literally made in the ’60s and, like, packed away in Siberia somewhere.” Unfortunately, his derision was predictive.
The US space agency had been proud of having two new space vehicles to service the ISS and was looking forward to the return of human spaceflight to US rocket ranges again. But now it had zero vehicles.
If SpaceX was out of commission for as long as Orbital, it would put great strain on the astronauts in orbit, who now had two fewer vehicles keeping them supplied with food and busy with research projects. (The space agencies in Japan and Russia also flew supplies to the station, but a Russian resupply machine had also recently failed, and it was no simple feat to add extra flights to the schedule.)
Bill Gerstenmaier, the top NASA spaceflight executive, told me that the time he spent preparing policymakers for potential SpaceX and Orbital failures helped them return to flight more quickly afterward. “I wanted to avoid what happens in my world, where we have the failure, the big investigation, it drags on for three years, we fix every problem,” he told me. “I couldn’t tolerate that with these cargo providers. I wanted them to fly absolutely as fast they could again. I wanted to let folks know we purposely went into this with high risk and we should expect failures.”
