By Ben Iannotta|February 2019
OneWeb Ltd.’s vision of spreading broadband communications around the globe will require mass producing hundreds of 150-kilogram satellites. A little over three years ago, OneWeb and Airbus established a separate joint venture, OneWeb Satellites, to make satellites for the main OneWeb company and someday other customers too. The first satellites produced by this venture are scheduled for launch this month on an Arianespace Soyuz rocket. The next challenge will be to accelerate production from the 10 satellites built so far to about two a day or 35 to 40 per month. This will be done at a newly built manufacturing facility in Florida. For details, I spoke to Tony Gingiss, CEO of OneWeb Satellites, during last month’s AIAA SciTech Forum.
IN HIS WORDS
Balancing risks of failure
… I gave a great analogy to someone yesterday. We were talking at dinner, and I said, “When drones first came out, private drones, I bought one for like $400.” OK, I’m an engineer, so I’m inherently probably a little thrifty in that I don’t want to break this expensive toy that I bought. I had the HD camera and everything, and you could fly it. I was always very cautious with it. I’d fly it and I had fun, but I didn’t want it to get too far away. I didn’t want to crash it into a tree. I didn’t want to get it stuck somewhere. It’s 400 bucks. I didn’t want to break it and have to rebuild it. It was fun, but it was never quite the fun that I thought it was. Then, someone bought me a $29 micro drone, a quadcopter little thing. No camera but a little controller. Twenty-nine bucks. I’m like a madman with it. I’m willing to take more risk. Every time I crashed I was thinking, “Oh. That’s $400.” It’s like, “Oh, if I crash, it comes with this little spare bag of parts that I bought for 6 bucks. If I really destroy it, it’s $29.” I have more fun with that thing, and I learned how to be a better drone pilot with that than I did with the $400 one with all the additional fancy technology, because my risk equation was different. I actually think that analogy applies to what we’re doing in the space industry today. I’ve worked on programs that took decades to build and cost things that didn’t have millions of dollars but had billions of dollars after them, and those weren’t zero-risk either. Nothing we do in our business — nothing you do in life — is zero-risk. It’s about managing risk. When you spend a billion dollars on something or $500 million on something and you’d spent 10 years of your life and government dollars or private dollars or people’s time, the risk you’re willing to take that something may be a step too far is going to be very different than if you spent sub-$10 million for one or two of them, and you just spent 12 months developing it.
Averting creation of orbital debris
We’ve put a lot of work into making sure that these satellites, ultimately, are de-orbitable. The highest reliability number we have for the satellite is de-orbit reliability. It’s the most important thing. Even higher than our mission availability. The avionics and the Hall-effect thruster propulsion system [have to work], and we have a very high reliability that we have for that. The UK Space Agency is kind of who OneWeb works through for getting all their launch licenses, so that’s something they’ve been very keenly aware of and interested in.
We actually designed-in what we call a grapple fixture. It’s essentially a fixture that ties into the main structure of the vehicle. It’s physically kind of a knob, if you will, that you can actually grab if you had a service recovery vehicle. You could actually go pull a dead satellite down. We’ve designed that in as a backup capability. Should we ever need that, there would always be a way to capture that satellite and bring it back down. That was designed to be tied right into the structure so it could withstand that pulling force, and not pull the satellite apart, which would be worse.
I’d say we’ve put a lot of thought into [de-orbiting] right upfront. Plus, we’ve put in kind of these backup plans should we ever need some capability like that. That [on-orbit servicing] industry doesn’t yet exist, but we’re trying to pre-position ourselves and our customers with hardware. That was proven hardware that have been used for things, so that as those capabilities evolve, they could use them if they ever needed to.
I think way early on, there was a lot of talk of a half-a-million-dollar satellite. OneWeb basically confirmed at EuroConsult that they’re paying under a million dollars for a satellite. And they’re buying 900 of them. That’s kind of the price point when you have a proven design, and we’ve gone through it, and you’re buying a very large quantity. I’ve always said if people want a price point, we’re talking about low-digit, single millions of dollars. Sub-$5 million for a satellite that’s, in some cases, an eighth the capacity of a geo communications satellite; a fraction of the weight, a pretty significant fraction of the comm capability, and one-one-hundredth the price, in some cases.
They’re right at about 150 [kilograms], but we’ve been working very closely with [OneWeb] because we want to give them the best capability to launch as many as possible. And if we can save a couple kilograms a satellite, that may mean the difference between them getting 36 on a vehicle versus 34 or 32.
Facilities in Florida and Toulouse
When we’re done with Florida, we will have three production lines. We’ll have one in Toulouse and two in Florida. The Florida facility is a ground-up facility. So in the design contract for OneWeb, we’ve qualified the satellite design. That’s what we’re finishing up right now. We built the first 10 vehicles, and we essentially designed and proved out the production system in Toulouse, France, right on the edge of the Airbus facility. We have our own building.
The Florida facility is actually a clone of that production system. We’ve proven it out over there. We clone it and basically make a times-two version of it. The Florida building is probably about four or five times larger than what it is in Toulouse. We were a little space constrained there because we were dealing with an existing building. The Florida facility is a ground-up facility with two complete factory assembly lines. That’s four module stations each with two full final assembly lines where we bring all the modules together and integrate vehicles, sized, basically, to do two satellites a day.
Assembly line commissioning
We’re actually going through what we call commissioning right now of the factory assembly line. We’re running real flight hardware through the first line. And the second line will be up about a month behind that. First line should be up and producing modules starting in March, and the second line will be up roughly a month later than that, producing module-level assemblies. Then we’ll have vehicles coming out starting a few months later starting all the way through the first batch.
We have four module stations. We have a propulsion module, which is basically the propulsion tank and the thruster. We have an avionics module, which is pretty much what you think of as the bus. It has all of the star trackers, and the wheels, and the flight computer, and that kind of thing. Most of it is on a single panel. Then we have a payload module, which is actually several panels. It’s kind of the front payload Ku antenna and the side payload panels. Then we have the solar array modules or the solar wings, essentially. So, all of those things together create kind of the web-shaped cube that you see as our vehicle, and every module is assembled, basically, in a single shift. Then final assembly of all of those together is less than an eight-hour shift. So, if you think every shift you’re producing one — and on a single factory assembly line — you’re producing one of every type module and integrating a whole vehicle.
We have test chambers. We call them Day-in-the-Life chambers, or diddle [DITL] chambers, where we do thermal cycling. We do that for a period of one to two weeks in that order. Then we have connectivity cabins where we actually power the whole spacecraft with light, as you test as you fly. Power them up as they would be in orbit, and actually do SOC [satellite operations center] connectivity with the customer side.
The scene in Florida
It’s pretty exciting. I mean Blue Origin was a little ahead of us in terms of build. I think we’ll go into production first, with building little smaller things in there. They’ve been great neighbors. They’re right across the street. At Space Florida a lot of companies have labs and stuff in that facility. They’re literally just half a block, a block from us. So, basically, in that one little area, there’s the big Space Florida building, us and Blue Origin.
We are hiring. Our big push now is we’re kind of finishing up the design phase. Florida is really focused on production, so we are hiring, basically, people to staff our factory. We will be hiring some engineers, but again, the near-term stuff is not really an engineering problem. It’s a production focus, so we’ll be hiring probably roughly 100 to 110 people just within the next six months. We’re right at about 170 people right now, where roughly 90 of them are in Toulouse and the rest of them are in Florida. But by the time we get into full production, we’ll be at basically 250 whereas Toulouse will kind of stabilize and maybe even go down as we exit the production phase at least in the near-term. Then, Florida will ramp up with staffing essentially two production lines. It’s about 90 people to staff both production lines at a single shift, kind of full staffing. So, just right there, you have roughly 100 people just staffing the floor.
Role of Toulouse vs. Florida
OneWeb is nominally slated to be all produced in the U.S. once we get done with the first 10. Production in Toulouse really will be for what we call third-party sales. So, Airbus has the third-party sales rights, and if they’re selling into Europe, we would probably produce over in Toulouse. If we’re selling the U.S., whether it be a [DARPA] Blackjack or some other commercial customer, we would ultimately probably produce that here in the U.S. at one of our two production lines here. So, we’ll maintain all three production lines as production. It’s really about: If we built it there for a customer there, why would we build it here and then ship it across the Atlantic? The other thing is we fully expect that we’ll have things for certain customers who will want it to be done in the U.S., and there will be certain things that certain customers in Europe will want to be done in Europe.
Satellites to date
By the time we get into February, we’ll be done with all 10 [required initially], and we’ll be ramping up Florida.
People ask me, “Oh, is the satellite cutting-edge?” I said, “If you just look at our satellite, there’s nothing about it that you say, ‘I’ve never seen this technology before.’” I would actually argue that, similarly, with our production system, there’s nothing that I would call bleeding-edge. We wanted proven technology. Things that we knew worked. I think that what’s really novel about what we’ve done is that if you look at the price point that we’ve established for our design, for the kind of quality that we’re providing, but the price point and the schedule for the design is unmatched. I mean no one’s ever done it for this, at this kind of production quantity; at the price point we’re at and at the speed.
We actually have a series of cameras at each of the module stations. As you go through, all of the work instructions we have are digital. We don’t have any paper on our floor. No logging or anything. That’s all done automatically. When the operators are done with certain steps, they can actually manipulate 3D models if they wanted to see a different view of what they’re doing. Many of our work instructions have visual models, where it’s actually showing them what to do. They’re looking at computer screens suspended right in front of their face above and off to the side of one of the module stations where they’re working. There’s a series of cameras that sweep across the module that they’re assembling, and it compares that to, essentially, CAD renditions, I’ll call it for a lack of a better term. It looks for: Are all the parts installed correctly? Is something backward? Is something missing? Does something look out of tolerance? And, you know, your goal is to get to 100 percent automated visual inspection. We’re not there. We’re probably into mid ’90s. In some cases, maybe higher. Some modules, it’s more complicated because there’s variations we have to account for, but we’ve achieved a really high level of automated visual inspection. It’ll flag things and say, “This is not right. Either something’s missing here, or something’s not installed correctly.” And it’ll actually flag it for the operator on the screen, show them where the problem is, and it will not let them continue until they resolve the issue.
Down on robots
People ask me: “Why haven’t we invested more heavily in robots?” I said we’ve looked at it, and for the dollars, it was cheaper to buy and have skilled technicians do certain things than to go invest in robots and go that way. They’re not as easy to adapt as we probably need them to be. The cost-benefit just wasn’t there for us. So, everything we look at is: 1) We want it proven. It’s not that we don’t want it leading-edge, but we don’t want it bleeding-edge. We want to make sure that they’re actually meeting our needs, and we’re very, very conscious about cost and schedule, and then maintaining quality at the end. So, it’s a fine balance.
Limited use of additive manufacturing
We actually don’t do a lot of additive manufacturing. We’ve used it in a couple of areas where we think for tooling or other things it’s helpful. It hasn’t been a big unlocking thing for us that it’s allowed us to do something we couldn’t do otherwise. It’s something we continue to look at because some suppliers will come back and say, “Hey, I could do the next generation of this with additive manufacturing and maybe bring the price down a little more or increase the volume of our production, or whatever.” So, I think we’re always looking at it as a tool, but in and of itself, it hasn’t been a thing that’s unlocked our ability to do what we’ve done.