by Tim Fernholz
As the rocket ignited, it seemed as if everyone watching was holding their breath or clenching their stomach until the rocket passed the moment where the CRS-7 vehicle had exploded mid-flight. This time, the struts—and everything else—held together. At stage separation, the eleven satellites were sent on their way. This was the primary mission, and in one sense this was all that mattered: They had proven that the Falcon 9 could be trusted to bring cargo to space again, and finally completed the build-out of a satellite constellation.
But for Musk and his engineers—not to mention all the fans watching on YouTube—the real prize would be bringing the booster rocket back down to earth. It was important enough that Musk personally delayed the mission by twenty-four hours because SpaceX’s computer simulations predicted a 10 percent higher chance of a successful landing.
Now it was all in the rocket’s hands as it flew itself with the software that Blackmore and his team had written, operating the fins and legs and valves that technicians had cleaned, tested, and double-checked ahead of the mission. Fifty miles above the Atlantic, the booster fired its engines to do a backflip and return to the Florida coast from which it had departed just four minutes before. Roughly eight minutes after liftoff, the engines restarted for a third time, to slow the rocket through atmosphere reentry. Two minutes later, it lit the sky around the Cape with orange fire and lowered itself to the ground with agonizing slowness, settling down wreathed in smoke. When it blew away, there was the twenty-story-tall machine, standing proudly. At SpaceX headquarters, the excited screams of the staff reached hysterical proportions. They had landed the rocket.
This was also a first: a vertical-takeoff rocket delivering cargo to orbit and then returning to land under control. In an industry that was afraid of failed tests, it’s hard to understate the chutzpah of returning to flight with a mission that tested three things: a brand-new rocket, a landing maneuver that had never been successfully demonstrated, and, after the Orbcomm satellites were deployed, the ability of the second stage of the rocket to ignite again. The latter was a crucial capability if SpaceX was to launch geostationary satellites.
Bezos couldn’t help but weigh in on SpaceX’s accomplishment. “Congrats SpaceX on landing Falcon’s suborbital booster stage,” he tweeted. “Welcome to the club!”
The sly reference to a suborbital booster stage couldn’t have been intended to do anything other than tweak Musk, who had been at pains to note that the Falcon 9 first stage could reach orbit on its own, even without the second stage. The SpaceX founder was busy racing to the landing zone to examine his booster and did not reply, but in the meantime the Amazon billionaire received a barrage of salty replies from Twitter users, accusing him of everything from jealousy to misunderstanding how rockets work. Regardless of where you come down on the more impressive achievement—being the first to land any rocket at all, or landing an orbital-class rocket—it’s clear that the two principals were fully enmeshed in a full-on reusable-rocket rivalry.
Bezos would take the next major step of actually reusing a landed booster. SpaceX brought the Orbcomm rocket back to its headquarters in Hawthorne, setting it up outside the main building as a totem and memento of what had been done so far. (The first Dragon space capsule was already hanging above the cafeteria.) Blue Origin, however, put its used booster back to work almost immediately. In 2016, it would fly and land the same vehicle four more times, gathering critical data about its flight profile and what it took to refurbish the vehicle for reflight. After each one, employees painted a tortoise, rearing proudly on its hind legs, on the New Shepard’s hatch, embodying the company’s “step-by-step, ferociously” motto. During the final test, in October, the Blue Origin team demonstrated the New Shepard’s abort system, proving that if something did go wrong with its rocket, the humans on top could jettison their capsule and fly to safety. The mid-flight abort involved shooting the capsule’s emergency rocket engines directly into the top of the booster, so it was not expected to survive. It did anyway, flying back to its pad in another proof of the resiliency of the hardware and the capacity of the flight software. In 2017, Bezos and his team would be awarded the prestigious Robert J. Collier Award, recognizing the New Shepard as “the greatest achievement in aeronautics or astronautics in America” during the preceding year.
After five successful flights, Blue retired the booster and entered into another one of its regular dormant periods: it would not launch another rocket for well over a year, focusing instead on more hardware development. “It looks to me that the original tests are what we call developmental tests,” one engineer with close ties to the company told me. “Now they are going to move into operational testing, which is what you do on the operational vehicle to make sure that you have a fleet you can start flying tourists on.”
Musk’s team weren’t resting on their laurels, either, and all of their tests made money for the company. The ground landing was all well and good, but for reusability to be useful in the long run, the company would need to perfect the seagoing landing scheme they had battled Bezos for in the courts. In January and March of 2016, the company attempted to land two more boosters on the drone ships. The first, a comparatively easy flight to low earth orbit, almost went swimmingly, but one of the landing legs came unlocked and the rocket slowly tipped over. The second flight was returning from a high-velocity mission to sling a satellite toward a high orbit; that rocket smacked into the ship with punishing force. Each time, the reusability team learned more about their vehicle, playing with different ways to balance engine thrust and maneuverability while conserving propellant.
Finally, in April 2016, SpaceX landed a booster on a drone ship following a mission to the International Space Station for NASA. Three more would follow suit after later launches, including two returning from high-velocity missions to geostationary orbits, and another would land on the ground pad. All in all, the company returned five boosters during that calendar year. The challenge was still in preserving enough margin from the tough job of going to space to get back down without losing control. SpaceX’s engineers focused on how to cram more power into their vehicle so they could make reusability financially sustainable.
Just as important as innovation for the company was reliability, expressed in the form of frequent launches: By the end of that summer, SpaceX had launched eight successful missions, each of which generated valuable revenue and useful data and cleared the decks for more jobs ahead. If it could launch just four more missions by the end of the year, it would match United Launch Alliance’s twelve missions for the year. To genuinely innovate and to equal an established incumbent for productivity would be a dual achievement for SpaceX.
That September, the company’s operations team at the Cape prepared for a launch on behalf of an Israeli satellite maker. SpaceX’s rockets usually go through a premission static fire, the procedure in which they burn through a full duration while held down by clamps—a dress rehearsal for space. Most companies would do this before putting the payload on top of the rocket. SpaceX, in its endless quest to save time and money, had already mounted the satellite in its protective fairing on top of the rocket. Pumps began loading fuel into the rocket ahead of the test.
Without warning, the Falcon 9 exploded. The rocket, the $175 million satellite on top, and the launchpad that SpaceX had developed at a cost of $25 million were all consumed by flames.
15
Rocket Billionaires
I know perfectly well that the hardheaded businessmen, who, after all, are really the ones who put research developments on a going basis, are convinced only by final accomplishments, and are not influenced by theories alone, however sound they may be.
—Robert Goddard
A failure during the intensity of launch is one thing; a mysterious explosion during a routine propellant top-off is a murkier problem. What could have happened to create such a mess? Conspiracy theories abounded, especially after SpaceX requested access to the rooftop of a facility operated by its ri
val United Launch Alliance that was within sight of the accident. The US Air Force would find nothing related to the fire during its inspection.
Once again, SpaceX was in scramble mode, trying to pin down what had gone wrong during the incident—once again classified as a mishap because no one was injured; safety procedures require a clear pad during fueling. The explosion—really just a “fast fire,” Musk mused online—had repercussions that signaled just how far SpaceX’s influence reached into the global economy.
An Israeli company called Spacecom had built the doomed satellite. Called Amos-6, it was designed to provide internet access to the Middle East and Africa. Eutelsat, a European telecommunications firm, had leased some of its capacity. More unusually, so had Facebook, the American social media giant. Facebook and its largest rival, Google, were focused on growth, but adding new users was becoming more about increasing basic internet access than convincing current netizens to use their services.
Just like the founders of Microsoft had before them in the 1990s, Facebook and Google began to look up at the sky. Google’s moonshot factory—the division where it developed risky futuristic businesses with its endless stream of advertising dollars—invested in high-altitude balloons and developed plans for a satellite constellation. Facebook’s strategy included solar-powered planes that could fly for days at a time, providing internet access to people below. But for right now, Amos-6 was a more straightforward way to boost internet access for markets in Africa. In booming metropolises like Lagos, Nigeria, and Nairobi, Kenya, broadband penetration was low, but people enthusiastically adopted mobile phones in their daily lives. Facebook’s effort to increase access was often pitched as philanthropy or corporate social responsibility, but there was real money at stake for the firm. Not just American but also European and Chinese companies were eager to win African consumers over to more data-intensive services—if the telecom infrastructure could be put in place.
The preflight destruction of Amos-6 was a dent in Facebook founder Mark Zuckerberg’s hopes to make a splash serving internet from space. On the day of the incident, he was making a surprise tour of African tech hubs—one whose timing on the day of the satellite launch was probably not coincidental. Writing from Nairobi, the internet billionaire said he was “deeply disappointed to hear that SpaceX’s launch failure destroyed our satellite that would have provided connectivity to so many entrepreneurs and everyone else across the continent.”
Disappointment reached further across the globe: When the Falcon 9 burst into flames, Spacecom was in negotiations to be purchased by a Chinese telecom firm for $285 million. The deal was contingent, in part, on the revenue from an operational Amos-6. Now the company would lose money, even after insurance, due to delays and replacement costs. Their merger was scotched—though China’s space ambitions were not. The burgeoning global power had sunk serious resources into its space program in recent decades, developing reliable launch vehicles and spacecraft, even if its approach to safety was what you might expect in an essentially authoritarian country. In 2011, it had launched its first temporary space lab, Tiangong-1. When Amos-6 was consumed by flames in September 2016, the Chinese were months away from sending two taikonauts to Taingong-2, a larger orbital lab.
These actions underscored the fact that the United States, the pioneers of human spaceflight, still did not have the ability to fly astronauts. The US government was still paying the Russians, all geopolitical tensions aside, to fly Americans, as well as their foreign partners from the European Union and Japan, up to the ISS. Russia had clearly noticed that NASA had no other options: a seat in the Soyuz capsule had once cost as little as $21 million, but by 2016 Roscosmos was planning to charge as much as $81 million. That was more than the price of a single Falcon 9 commercial launch. The total cost of relying on the Russians between 2006 and 2018 was forecast to be as much as $3.4 billion. The financial pressure, plus humiliation as other nations pushed past them in space, fell hard on NASA and the two companies racing to get Americans back up to the space station through the Commercial Crew program: SpaceX and Boeing. Neither was on schedule.
Having to spend another six months figuring out what had gone wrong with its rocket wasn’t going to make things any easier for SpaceX. The company quickly zeroed in on the source of the problem. To increase the power of the Falcon 9 second stage, SpaceX engineers were super-chilling the liquid oxygen (LOX) that combusted along with the rocket fuel down to minus 340 degrees Fahrenheit. This made it denser and allowed them to pack more into the tank. Musk said that using the superchilled fuel was key to “full reusability” of the rockets. If the second stage could fly further on its own, the booster could conserve more fuel for successful landing attempts. In order for the LOX to remain cold ahead of takeoff, the ground crew would begin filling the tanks just thirty minutes prior to flight, a procedure known as “load and go.”
SpaceX had experimented with different techniques for loading propellant into the Falcon 9 throughout 2016. This led to launches being scrubbed at the last minute when pressures or temperatures were outside nominal levels. It was another shared challenge faced by rocket companies pushing the edges of performance: the technicians behind the computer-controlled plumbing that precisely fueled the Falcon 9 were distant cousins of the X Prize–winning flight engineers who awoke at 2 a.m. to slosh SpaceShipOne’s nitrous oxide around in a tank until it reached the correct temperature. Trial and error eventually left SpaceX with a process that, it seemed, could reliably load the superchilled propellant.
Outside observers were fretting about this approach when it came to SpaceX’s plan to fly humans. A NASA advisory board led by retired astronaut Thomas Stafford, who had flown Apollo 10 around the moon, wrote a warning letter to Bill Gerstenmaier, saying that “load and go” would be unacceptable when SpaceX started flying with people on board. “There is a unanimous, and strong, feeling by the committee that scheduling the crew to be on board the Dragon spacecraft prior to loading oxidizer into the rocket is contrary to booster safety criteria that has been in place for over 50 years . . . Only after the booster is fully fueled and stabilized are the few essential people allowed near it,” Stafford wrote. This wasn’t entirely true. As SpaceX engineers pointed out to me, the space shuttle, though fueled before the crew boarded, had its propellant tanks constantly replenished until just minutes prior to ignition as liquid hydrogen and oxygen boiled off. After the Amos-6 accident, a different NASA safety panel warned that analyzing the superchilled fueling system would not be a “trivial effort” and that NASA shouldn’t let concerns about budget or schedule force it to rush into using a novel technology that it didn’t completely understand. “Systems often display ‘emergent’ behavior once they are used in the actual operational environment,” the advisers noted drily.
In the months after the mishap, SpaceX was true to its experimental ethos and began attempting to replicate the accident at its McGregor test facility. By the end of October, it had a theory about what had happened. It centered on carbon-fiber tanks known as composite overwrapped pressure vessels, or COPVs. Each COPV had an aluminum liner and held helium that, during flight, was pumped into the propulsion system to keep LOX flowing into the engines at high pressure. During the disastrous CRS-7 flight in 2015, it was one of the helium tanks that had broken loose when the strut holding it in place snapped. Investigators were able to rule out the same fault in this case.
SpaceX was pushing the limits of COPV technology. Space engineers had long known that the carbon composites could interact explosively with liquid oxygen under the wrong conditions. Ultimately, SpaceX’s experiments found that the superchilled LOX in the Falcon 9 had soaked into the composite wrapping. In some cases, the oxygen grew cold enough to actually change its physical state, from a liquid to a solid. During loading, as the helium filled the COPV, the slushy oxygen pooled in small dents or deformities in the aluminum liner, known as buckles. Semisolid oxygen collecting in the buckles could be forced against the woven fibers of the
composite; if they cracked or rubbed together, a spark could—and apparently did—ignite the entire rocket.
“The reason that that happened was because there was a big push to compress the timelines and go faster and faster and faster,” a former SpaceX employee told me. Less time on the pad meant cheaper and faster launches, and the company hoped to launch a rocket within an hour. That required pumping liquid helium, rather than warmer compressed gas, into the rocket to fill the tanks more quickly. “They just learned some lessons that they hadn’t discovered through any kind of test program in Texas. They loaded fast enough, they had a problem with the rocket—boom.”
Failed experiments with composite tanks had been a key reason for NASA’s canceling a prototype replacement for the space shuttle—the Lockheed Martin X-33—in 2000. Pushing the envelope is a risky business, however, and as SpaceX worked to wring every ounce of power out of its vehicles to reach its goal of reusability, its margin for error declined. Nonetheless, after the expensive accident, SpaceX didn’t abandon the program. Instead it announced that it would revert to a helium-loading procedure that had worked more than seven hundred times, and redesign the COPVs to prevent the buckles.
The accident had done significant damage to the launchpad, SLC-40, and that would require time and money to repair, leaving the company temporarily without an East Coast launch facility; they were still refurbishing the second launchpad, SLC-39A, leased from NASA three years before. Just as in 2015, SpaceX had been aiming to increase its launch cadence and beat ULA for most American orbital launches. And once again, a public and embarrassing failure had left SpaceX behind the eight ball, though it had managed to launch two more rockets than in the prior year.
