The chance that SpaceX’s planned Starlink satellite constellation will cause an injury or death is 45 percent every six years, according to an IEEE Spectrum analysis of figures submitted by the company to the U.S. Federal Communications Commission.
Elon Musk hopes the nearly 12,000 satellites in the constellation will eventually carry half of all Internet traffic. The satellites will use laser and radio links to provide fast, cheap Internet access to people all over the world—and the associated service fees could help Musk fund his dream of colonizing Mars.
But what goes up, must come down, and the sheer scale of SpaceX’s orbital network means that it may only be a matter of time before the world sees the first injury or death from a plummeting satellite.
Starlink will actually be made up of satellites in three distinct orbital shells, each operating at different altitudes. The first to launch will be 1,584 satellites in low earth orbits (LEO) of 550 kilometers above ground. Following these will be up to 2,825 satellites placed higher up, orbiting at altitudes between 1,100 and 1,325 km; and then 7,518 satellites positioned lower down, in very low earth orbits (VLEO) centered around 340 km. The combined mass of the satellites will be more than 10 times that of the International Space Station.
The 11,927 satellites will carry solar panels, batteries, radio and laser communications gear, and enough fuel to adjust their orbits or counteract drag from faint wisps of atmosphere. When that fuel runs low, after around six years, the satellites will use the remainder to nudge themselves into the denser atmosphere below. Each satellite, just a little smaller in size than a Tesla Model 3, will then swiftly burn up.
Except that some of it won’t. SpaceX estimates that several kilograms of each 386-kilogram Starlink could reach the Earth’s surface with sufficient energy to harm or kill someone. NASA has fixed this figure at 15 joules—about the same wallop as a baseball traveling at 51 kilometers per hour. Depending on the satellite’s configuration, iron thruster components, stainless steel reaction wheels, or silicon carbide mirrors could survive the journey from orbit to your head.
When SpaceX plugged the numbers into NASA’s Debris Assessment Software, the package calculated that there was, at most, a 1 in 18,200 chance that an individual satellite in its LEO orbital shells would hurt or kill someone. VLEO satellites were generally slightly riskier, with up to a 1 in 17,400 chance. All figures are handily less than the 1 in 10,000 figure that NASA has adopted as a standard, and that U.S. and European space agencies require for space missions.
However, the FCC didn’t stop there. In March and June 2017, the FCC calculated the aggregate risk to humans from the entire constellation. Assuming the 11,927 satellites are launched on a regular basis, they will fail in the same way. Starting around six years from the first launch, an average of five satellites a day will reenter the Earth’s atmosphere, each with a tiny chance of failing to completely burn up, resulting in a part that could hit someone.
But with more than a thousand satellites falling a year, those tiny risks add up. The FCC figured out that, over their lifetime, satellites in the LEO shells posed a 1 in 5 risk of hurting or killing someone, and the VLEO satellites carried a 1 in 4 risk. IEEE Spectrum’s calculations using SpaceX’s most up-to-date information suggests that the overall risk of debris from the constellation causing an injury or death will be 45 percent.
This means that NASA’s software says that it is nearly as likely than not, that one of the Starlink satellites will injure or kill someone, about every six years.
Given that there are 7.5 billion people on the planet, the risk to any one individual is still extraordinarily low. When a satellite tumbled to the Earth in 2011, NASA calculated its odds of hitting someone to be about 1 in 3,200. But as Jason Samenow at The Washington Post pointed out at the time, an individual person’s risk of being hit was 1 in 22 trillion—many times smaller than their chances of winning the lottery or being struck by lightning.
Still, other space companies are concerned about the problems that any such incident, however unlikely, could create for their own plans. In September 2017, space Internet rival OneWeb pointed out that some components from SpaceX’s satellites would survive reentry with an energy of more than 960 joules—equivalent to a baseball travelling at more than 400 kph. It asked the FCC to “require SpaceX to address the aggregate risk posed by its very large LEO and VLEO constellations and adopt adequate measures to mitigate that risk.”
Of course, Earth is also constantly being bombarded by natural objects from space. In its response, SpaceX noted that between 62 and 242 meteorites larger than 10 grams make it to ground level on an average day. “Thus, although the SpaceX system may seem to involve a substantial number of new objects, it remains insignificant compared to the large number of meteorites that strike the Earth naturally each year,” wrote the company.
The FCC disagreed. In its letters to SpaceX, the FCC had calculated that falling Starlink satellites would result in as many as 500,000 separate objects reaching Earth’s surface over the six-year period. With Starlink satellites expected to last for six years, that corresponds to an average of 228 objects a day—more than natural meteorite impacts on all but the busiest days.
SpaceX responded by saying its satellites would create, at most, 10 fragments each over its lifetime, only a fraction of which would pose a risk to people. That works out to around 120,000 objects overall or 54 daily—comparable to a quiet day of natural meteorites, but hardly insignificant.
With all these extraterrestrial objects hurtling down around us, the number of recorded human injuries and deaths from space debris has been remarkably small. In 1997, Lottie Williams of Tulsa, Okla., received a glancing blow from what was probably part of Delta rocket burning up above her. And in 2016, Indian authorities claimed that a bus driver in the state of Tamil Nadu was killed by the blast from a meteorite strike. For comparison, an asteroid that exploded above the city of Chelyabinsk, Russia in 2013 created a shock wave that injured around 1,600 people.
SpaceX pointed out that most of us are not wandering outside waiting to be hit by celestial bodies. “Given that the DAS analysis assumes that debris with as little as 15 joules will result in human casualty, any level of shelter would significantly reduce the likelihood of injury,” it wrote to the FCC (NASA defines “casualty” as any injury to a human, including death). “Although SpaceX has not undertaken an analysis that takes sheltering into consideration—since it already complies with U.S. and international standards—it is worth noting that this factor alone was sufficient to reduce the casualty risk of another [satellite] operator by more than 400 percent.”
The satellite in that case, an Iridium NEXT, was found by the DAS software to have a risk of causing an injury or death as high as 1 in 4,400. Iridium argued that because only 10 to 20 percent of the world’s population are outside and completely unsheltered at any one time, the rest of us should be excluded from any risk calculation. The FCC authorized the NEXT satellites for launch.
SpaceX also promised to further refine its satellites’ component geometries to maximize the probability of them burning up on reentry, and to collaborate with NASA on designs that would minimize the risk of harming people. In November 2017, it signed a Space Act Agreement with NASA to use the agency’s advanced Orbital Debris Object Reentry Survival Analysis Tool software to analyze its satellites’ survivability in more depth. The company was also granted permission by the FCC to launch and operate its Starlink constellation, most recently in November.
But that does not mean the government has stopped worrying about the sky falling. On November 19, the FCC released a notice of proposed rulemaking on the subject of orbital debris. In it, the agency invited comment “on whether, when assessing human casualty risk, we should do so on an aggregate, system-wide basis as well as on a per-satellite basis, and, if so, what metric should be used to evaluate aggregate risk.” Comments are due early in 2019.
With SpaceX having launched only two Starlink prototypes to date, there is still time for regulations to have an impact on the constellation’s future. But not much time. Under FCC rules, SpaceX must have half of its 11,927 Starlink satellites in orbit and operating no later than November 2024. The company plans to begin launches in 2019.