Comment by davidcuddeback | Hacker News Reader

Comment by davidcuddeback | Hacker News Reader

Parent Full thread

> This is going to be the killer [NPI] feature of using automated rather than manned systems.

Using automated systems rather than manned systems is a necessary but not sufficient condition to allow for relaxed reliability requirements in spacecraft. Another necessary condition is reduced cost. It's common for a single mission to cost hundreds of millions of dollars, which makes many spacecraft designers very risk-adverse.

When I was in college, I worked at PolySat [1] where I wrote embedded software for CubeSat-class satellites. CubeSats are fairly revolutionary in the spacecraft business, because they can be launched for less than one million dollars (including development costs). The cost of the hardware was less than $10k for a single satellite and the launch costs are around $40k per kilogram (which is the maximum weight of a 1U CubeSat). Those costs are low enough that we built backup flight units in addition to engineering units. In one case, we were able to launch a backup flight unit of a satellite that was destroyed in a launch failure. (CP2 was on the crashed Dnepr-1 of July 2006 [2]. We relaunched the backup flight model as CP4.) That was only possible because the cost was so low. $50k to relaunch a satellite is unheard of.

[1] http://polysat.calpoly.edu/ [2] http://en.wikipedia.org/wiki/Dnepr-1

I agree with you, but once there are orbital factories (and how far are we from that with companies like this? 20 years?), the cost to "launch" becomes next to nothing.

>How Billionaire Asteroid Miners Make Money -- Without Mining Asteroids.

They're going to make a movie about this starring Bruce Willis. Throw in Ben Affleck for comic relief. They're going to call it "Armageddon" and it's going to be a blockbuster.

Twenty years seems a little optimistic. We don't have ground-based automated factories yet -- not fully automated anyway. And the cost to lift a ground-based factory (or even its parts) to space on tonnage alone would be astronomical. Plus, many things are harder in a vacuum. So you're looking at some significant technological progress before what you're suggesting will be possible.

Let's target rational exuberance.

Automated factories are coming from both ends - continuous production, and custom on-demand stuff. Mines / oil rigs / pencil factories are getting almost entirely automated, because they never have to change what they are doing. 3D printers are becoming automated, for the opposite reason. Being squeezed from both sides, non-automated manufacturing is not here to stay.

I'd be more pessimistic on the grounds that there isn't a huge market for space-based factories. Making the best parts requires heavy machinery (whether or not it's automated), which you can't affordably launch into space.

I would imagine that it would be quite difficult to design a 3D printer that works in zero gravity. All current designs that I'm aware of rely on gravity to either keep the base material on a level surface (e.g. SLS, SLM) or to draw the build material down from a nozzle of some sort (e.g. FDM)

Repraps print fine on a vertical surface, or upside down. The inter-layer adhesion is achieved by melting the new layer into the previous one, and the build material is pushed down with positive pressure. It would work fine in zero gravity, though cooling might be a problem if it were working in a vacuum.

Everyone forgets about 3D printing's older brother, 3D milling. Start with a big block of aluminum (or any metal softer than the cutting bit), use what amounts to a CNC dremel tool to carve away everything you don't want. It is vastly simpler than 3D printing and creates objects that are much more sturdy. But it is slower and uses a more expensive feedstock than printing.

Good match for space fabrication, too. Zero-g makes it much easier for an air nozzle to dislodge debris. Feed it big ingots of metal straight from the asteroid smelters. Not sure if you could do the milling in a vacuum, but it would be worth a shot. Downside, greatly limits cooling. Upside, easier recovery/purification/recycling of shavings. A big electrostatic charge on the ingot and tools should make the shavings literally fly off of the work to a collection plate.

It's even better than that, right? In microgravity, what's stopping you from sculpting a precise 3D shape out of a lump of molten metal?

Do you mean having a blob of molten metal floating in a chamber? Surface tension means it won't keep a shape. (I guess you could make cheap ball bearings.) If you get around that, any time you touch the blob it would start rippling and ripples won't really damp out. Might make for some pretty sculpture if you can flash freeze it, but nothing of industrial value.

Unless you are trying to obliquely refer to casting, which should be pretty much identical with or without gravity/air. And still needs tooling to make the molds.

(And please don't say that the molten metal could be a feedstock for a printer. That would be a circular discussion.)

Um, but metal doesn't just freeze, right? There's a stage of plasticity while it cools during which you could feasibly sculpt with it.

Not saying it would necessarily be cost-effective, but it's an interesting idea.

Hey, you were the one who said molten :-)

I guess you meant something more like blacksmithing. Chamber could add heat by microwave induction. No air to oxidize the surface so you might be able to work trickier metals. We already have power hammers, we already have 6DOF manipulation tables. I'm not entirely convinced that a power hammer could do the work accurately enough, might want to use a milling machine for the final fitting.

I think that's going to be one of those things that's "about 20 years from now" for at least the next 50 to 100 years, like AI which automated manufacturing like this probably will depend on.

Orbital factories do not need AI; the relevant cost is the transportation of manufactured goods, a cost alleviated by remote controlled manufacturing.

Since the original article mentioned redundancy and planning for failure, I think we have to factor in the cost of parts when discussing launch costs. After all, it's the cost to launch a replacement. (Notice my $50k figure included the cost of parts for our backup flight unit plus the launch costs.)

I think you're right that automated factories in space will reduce the actual launch costs (just the cost of getting your junk into space). The rest of the cost reduction has to come from reducing the cost of manufacturing. Accepting additional risk can certainly help here. That's a big change of mindset for the aerospace industry.

In my example, the cost of materials and manufacturing was a fraction of the actual launch cost. I don't think that's typically the case for traditional spacecraft, but I'm not positive about that. Adding more or heavier parts increases the mass, which requires more energy from the launch vehicle to reach orbit. I find it entirely plausible that adding redundancy increases launch costs as well as manufacturing costs, but I don't know how the big boys arrange their launch costs. (At PolySat, our satellites were so small that we paid per kilogram. Other satellites can be the size of a van or even a school bus.)

I think 20 years is a bit optimistic for space-based factories to reach the point where we can dramatically lower the costs of launching a spacecraft. I think a kilogram of aluminum mined in space and a kilogram of aluminum mined on Earth would not be fungible, and would thus have very different costs. Initially, I think the aluminum available at the space-based factory would cost significantly more than ground-based aluminum. (The company that builds the factories will have to recoup their costs, as will the companies that mine the raw materials. And at least initially, they're going to be paying to launch those spacecraft from Earth.) Eventually, we could reach a scale that the cost of materials in space is low enough to seriously discount traditional launch costs. I just think there's a lot more than 20 years of work to get there.

Reply on news.ycombinator.com