The Space Reviewin association with SpaceNews

CubeSat launch
A group of CubeSats being launched from the International Space Station. While interest in CubeSats is growing among many different users, reliability of such spacecraft is a concern. (credit: NASA)

CubeSats: faster and cheaper, but better?

Bookmark and Share

Every year, it seems, interest in CubeSats grows. One way of measuring that is at the annual summer workshop for CubeSat developers, held earlier this month at Utah State University the weekend before the main Conference on Small Satellites there. The first workshop, held a decade ago, attracted a few dozen people who filled a classroom. This year’s workshop, now fully integrated into the main conference, had nearly 1,000 people attend, enough to move the sessions into the ballroom in the student union building that hosts the regular conference sessions.

Nearly one in five CubeSats suffer immediate problems so severe they can’t provide any telemetry about what might have gone wrong. “That’s way too high,” Langer said.

More people are attending the workshop because more people are building and launching CubeSats. Universities, pioneers in developing the CubeSat standard more than 15 years ago, continue to use them for science, technology development, and education. Government agencies are using them for various roles, including future missions beyond Earth orbit. And companies like Planet and Spire have built businesses, and raised hundreds of millions of dollars of venture funding, on plans to launch constellations of hundreds of CubeSats to take images, track ships, and collect weather data.

CubeSats have attracted that interest as technology makes those small spacecraft, particularly in 3U (10 x 10 x 30 centimeters) and 6U (10 x 20 x 30 centimeters) versions, capable of doing missions that once required larger spacecraft. CubeSats can be built quickly, as Planet has demonstrated with more than a dozen iterations of its 3U spacecraft built over just a few years. Not surprisingly, they are also much cheaper than larger spacecraft as well.

But are CubeSats better than bigger spacecraft? That, of course, is a subjective judgment that depends on how one defines “better.” By some metrics, though, CubeSats may be falling short of larger spacecraft, enough to worry some in the field.

In a talk at the smallsat conference August 10, Martin Langer of the Institute of Astronautics at the Technical University of Munich presented the results of an analysis of failures of CubeSats. Langer compiled a database of 178 CubeSats launched through the middle of 2014 (excluding those lost in launch failures), compiling information about when they failed and how.

Langer found that 18 percent of those CubeSats were “dead on arrival”: that is, they never made contact with the ground after deployment in orbit. In other words, nearly one in five CubeSats suffer immediate problems so severe they can’t provide any telemetry about what might have gone wrong. “That’s way too high,” he said.

For all CubeSats in the database, Langer said a third of failures have no known cause. That includes both the dead-on-arrival CubeSats as well as others that failed later in their missions. “That’s a problem,” he said.

“Should we be doing better?” Swartwout asked after presenting those statistics. “Probably we should be doing better.”

For CubeSats that operate for at least 30 days before failing, there is more data available on what caused the failures. Issues with the spacecraft’s electrical power systems account for nearly half of those failures, followed by on-board computers and communications. Only 12 percent of failures have no known cause. For spacecraft that last at least 90 days, the statistics are similar, with slightly more communications failures and slightly fewer power failures.

That’s not to say all the CubeSats in his sample failed: through two years, about 60 percent of the satellites were still working. And some CubeSats are remarkably long-lived. In another presentation at the conference, Shinichi Nakasuka of the University of Tokyo said that its CubeSat XI-IV satellite, launched in 2003, was still working and returning images of the Earth 13 years later.

However, that high failure rate, particularly the infant mortality rate for newly-launched CubeSats, is a cause for concern. One possibility is that the statistics are skewed towards university CubeSats, which may be more prone to failures than those built in government or business environments. Langer did not disclose the demographics of the satellites in his database.

In a separate presentation at the conference August 11, Michael Swartwout of Saint Louis University presented an update on “university-class” satellites, which includes larger smallsats but is dominated today by CubeSats. Those numbers have grown in the last few years as CubeSats became more popular, with about 30 to 40 university-built smallsats being launched a year in the last few years.

Swartwout examined 244 university-class smallsats launched from 2000 through 2015. About one-eighth of those were lost in launch failures. Taking into account just those that successfully made it into orbit, nearly 30 percent were dead on arrival, and an additional 15 percent failed early in their missions. Only about one quarter completed their full mission.

“Should we be doing better?” he asked after presenting those statistics. “Probably we should be doing better.”

Those statistics vary widely among different types of universities based on resources and experience. Among so-called “flagship” universities, which typically have access to greater resources and develop satellites more frequently, only about one quarter of their satellites were dead on arrival or failed early. By contrast, “independent” universities that had launched three or fewer smallsats had more than three fifths of their satellites be dead on arrival or fail early.

“In CubeSats, we have to accept failure as an option,” Langer said. “We have to learn from that.”

Despite those high failure rates, Swartwout said that university-class smallsat missions do matter. “They are producing publishable data, science data that can go into reputable journals,” he said. “They are doing technology infusion.” He added that they also “reduce the fear factor” for trying riskier missions in the future.

But while university CubeSat and other smallsat missions can perform useful missions, he offered a mixed assessment about whether they were worthwhile for an individual school. Of the 132 universities worldwide that have flown smallsats, he said, 80 have done it only once. “It takes a lot of resources, a lot of investment, to fly that first one, and then you wonder if it’s worth it,” he said.

Swartwout recommended universities developing smallsats do more to share lessons learned to improve the chances of success. “There is value in listening to people who have done this before,” he said.

Langer had similar advice about improving reliability. “In my opinion it’s not possible to achieve it using the traditional mission assurance. That’s way too big,” he said. “We can’t do that with CubeSats. We have to be fast, we have to learn things, we have to use new components.” He was similarly skeptical about the use of “high-reliability” components traditionally used by larger, and more expensive, spacecraft.

He recommended greater use of system-level testing of CubeSats prior to launch to identify issues. He also called for more sharing of data and lessons learned among CubeSat developers, noting the difficulties he experienced tracking down information about failures for his database. “We have to provide our lessons learned to the community,” he said. “It would be great to share more lessons learned in the future.”

That might be harder and not easier in the future, though. A growing number of CubeSats are being developed by companies, who may be less willing to share technical details about the failures of their satellites as they seek competitive advantages. Langer said he excluded CubeSats built by Planet from his database entirely given the lack of information about what failures they experienced. And Planet is just one of several companies planning constellations of hundreds of CubeSats.

It may be in the overall CubeSat community’s best interest, though, to share at least some details about failures, perhaps to help identify particular components susceptible to malfunction. A forecast by SpaceWorks published earlier this year and presented at the conference projects up to 3,000 satellites weighing 1 to 50 kilograms, dominated by CubeSats, could be launched through the early 2020s. A continued high failure rate will only add to the perception held by some, particularly among government agencies and companies that operate larger spacecraft, that CubeSats are “space junk” that pose a growing risk to their satellites.

But even with greater information sharing, Langer argues that CubeSats will still suffer failures. “In CubeSats, we have to accept failure as an option,” he said. “We have to learn from that. We have to make failures, and we have to provide our lessons learned to the community.”