The 3D printing of Lunar Base Shackelton.

[Mazon Del]

You're talking of building houses out of wood, which is pretty done only in North America nowadays, the rest of the world preferring concrete or bricks.

Typical cheap houses are made of cement blocks or prefabricated cement walls. The quality is not very good, insulation requires additional materials, and you end with square boxes. A concrete printer would make solid concrete walls (long lasting) with no restriction in shape, with very little human work. It can make walls with lots of empty cells for insulation, prepare ways for cables and pipes, do stairs, domes, pizza ovens and all sorts of wacky shapes with no additional costs.

The thing will probably not be very cost effective for single houses for a long time, but will be a great tool if you want to build 50 or 60 houses in the same place. I can also imagine applications beyond houses, like industrial or military structures.

And remember, once you have the machine, it costs next to nothing compared to human workers.

Exactly! The intended result of the machine I was discussing was that you'd have a line of rails running parellel with the road you are building on. Once the printer has finished with one house it moves to the next. The only thing you need humans for is to keep it supplied with material, to make sure it isn't having any issues (it's a machine, machines have issues from time to time, same as people), and to do the detailing work after its done.

[dansmithers]

It can make walls with lots of empty cells for insulation, prepare ways for cables and pipes, do stairs, domes, pizza ovens and all sorts of wacky shapes with no additional costs.

Since when? In the video it lays down a 2' by 4' bead of concrete. Not exactly precise. That's what happens when you scale a 3D printer: it's either very, very slow or ridiculously imprecise.

This idea is an outlier, anyway. Most of what I was harping on was people saying "In the future, we could 3D print airplanes/cars/guns/any device with more than 3 moving parts."

[Mazon Del]

Since when? In the video it lays down a 2' by 4' bead of concrete. Not exactly precise. That's what happens when you scale a 3D printer: it's either very, very slow or ridiculously imprecise.

This idea is an outlier, anyway. Most of what I was harping on was people saying "In the future, we could 3D print airplanes/cars/guns/any device with more than 3 moving parts."

The nozzle width is fixed at what it is. But that doesn't mean it is not precise. If they can position the edge to a micron (0.1 mm, or high resolution for current printers), then as long as you don't require a feature that is less than the width of a single nozzle path, you are fine. Now in the video of the prototype, they were caring more about proving that the overall concept design rather than on improvements that make the final one better.

I hate to be the stereotypical example of what you hate...but we WILL be able to do that. We can already do that. I can 3D print an entire clock right now if I wanted to as one piece requiring no assembly from me. If I went really fancy it could even have the chain in place, then I either print off a heavy weight, or I just shove a tiny block of metal into the weight bucket. Bam! Clock. If I want, I could design it so that all I have to do is slot in a DC motor with a battery pack and everything else is finished.

Frankly right now if Boeing wanted to, they could likely print off the engines to a 747 or whatever as single print jobs, then slap on the assorted electronics and whatnot. They don't WANT to do that mostly because they don't get anything out of it. Right now to switch their tooling over would be exceedingly costly because of how heavily invested they are in their current mode of production. Additionally, at this time they do not particularly gain anything out of doing so. But eventually once a printer exists that lets them produce the engine with all the electronics and whatnot on board the calculus will change. If suddenly it takes a month to build one engine, but they work on 4 in the space they normally would work on 8 AND they don't have to pay hardly anyone for the process. Even if their usual build time is two weeks, they will have a hard time justifying NOT doing this. After all, for an initial massive expenditure, they will be able to match their production rate and then over the course of several years enjoy the fact that they went from many hundreds of workers down to like 20-50 people.

Now of course, the big question is, is there anything that bulk manufacturing can still do to keep the balance tipped in its favor. The answer is certainly. But you will see some markets end up dominated by 3D printed production. Particularly in the realm of specialty device markets. Companies HATE to do 1 time engineering if they don't have to because it costs so much for so little gain. But with full on printing factories it suddenly becomes cheap to produce the prototypes or one-off runs of a device. Plus if suddenly the device manages to form a market around itself unexpectedly, they can just keep running off prints of it without having to devote the funds towards tooling up an assembly line. Look at all the Kickstarters that spend months and months sitting in limbo because they are working with the manufacturers on setting up production lines. Admittedly some of this can be solved by better planning on their part. But suddenly now they don't need to bother.

Right now we are sort of in a 'phase 1' of 3D printing, where we are getting better at any given material by itself (right now we can basically print metal by itself, or plastic, but not together). 'phase 2' is where we are working on printing multiple materials at the same time (metal on plastic, so basically doing circuit boards). 'phase 3' is either where we primarily work on increased speed and/or precision (printing the processors into the boards as the boards are printed).

[Idobox]

Since when? In the video it lays down a 2' by 4' bead of concrete. Not exactly precise. That's what happens when you scale a 3D printer: it's either very, very slow or ridiculously imprecise.

This idea is an outlier, anyway. Most of what I was harping on was people saying "In the future, we could 3D print airplanes/cars/guns/any device with more than 3 moving parts."

Concrete 3D printers are a very recent concept, with maybe one or two prototypes in the world right now. Of course, it's going to take some time before they are ready to be deployed. Think of what happened to others technologies over the past 30 years, like printers, 3d printers, digital cameras, drones, etc...

There is no reason a printer has to trade off speed and accuracy. With a smaller injection head, you can make finer details, which would indeed make construction longer if you used only one head. Look at a inkjet printer, and you'll notice they use multiple injectors in an array. Most consumer 3D printers have only one injection head, for cost reasons, but pro stuff, especially for powder beds, use a large number of injection heads.

And even if printing was slow, who cares? The machine works 24/24, costs pennies in electricity, and would likely give better results (once again, not current stuff, 5 or 10 years down the road).

Finally, there is always the option of using extrusion moulds, the same kind that is used to make hollow bricks, to directly print "patterns" in a cheap and dirty way.

The first uses will be for places where putting humans is dangerous or expensive, like space, disaster sites, war zones, deserts, underwater, etc... As the technology gets cheaper and better, it will be used in less extreme conditions, until you can get one for 200$ a day at every tool rental place.

[Idobox]

Frankly right now if Boeing wanted to, they could likely print off the engines to a 747 or whatever as single print jobs, then slap on the assorted electronics and whatnot. They don't WANT to do that mostly because they don't get anything out of it. Right now to switch their tooling over would be exceedingly costly because of how heavily invested they are in their current mode of production. Additionally, at this time they do not particularly gain anything out of doing so. But eventually once a printer exists that lets them produce the engine with all the electronics and whatnot on board the calculus will change. If suddenly it takes a month to build one engine, but they work on 4 in the space they normally would work on 8 AND they don't have to pay hardly anyone for the process. Even if their usual build time is two weeks, they will have a hard time justifying NOT doing this. After all, for an initial massive expenditure, they will be able to match their production rate and then over the course of several years enjoy the fact that they went from many hundreds of workers down to like 20-50 people.

I have to disagree with you here.

3D printing does not give the same quality as other techniques, and never will. The resolution you get is worse than the resolution used to make the injection head, so the technique used to build the printer will always be better than the printer.

And 3d printed metal is basically crap. Making a metal piece is not as simple as it sounds, you have to think of grain size, thermal stress, surface treatment, and a million other things. 3D printed metal is either metal powder held together by glue, or sintered which means it's porous, fragile, has a huge surface area, and is unsuitable for many surface treatments.

And by the way, Boeing doesn't build engines, they buy them, as does Airbus and other airplane companies. It is a very different job to make a plane and to make a turboreactor, one that GE or Rolls Royce has been studying for the past 60 years.

[Kerbart]

I'm sure this problem can be overcome, for example by some form of anchoring the excavator to the ground while it digs, but it's by far not as simple as you think it is.

Conceptually equip the excavator with buckets that get filled with moon rocks on site.

Also, earth excavators are built to excavate as much material as possible in the shortest amount of time. A lunar excavator will be designed around a different set of contraints and the time factor will be less relevant (or even completely irrelevant if a robotic excavator is used). Perhaps a portal-style design with the excavating bucket going down vertically in between the two axles? 3D printing might be the way to go, but excavating does not have to be so complicated, and there are a lot of advantages to building underneath the surface (protection from radiation for instance)

[Aghanim]

This is not 3D printing, this is burying an inflated preform with lunar regolith by automated excavator

[andrewas]

I hate to be the stereotypical example of what you hate...but we WILL be able to do that. We can already do that. I can 3D print an entire clock right now if I wanted to as one piece requiring no assembly from me. If I went really fancy it could even have the chain in place, then I either print off a heavy weight, or I just shove a tiny block of metal into the weight bucket. Bam! Clock. If I want, I could design it so that all I have to do is slot in a DC motor with a battery pack and everything else is finished.

IIRC, those printed pendulum clocks lose a couple of seconds and hour and run for less than a day. They're an interesting exercise, but not a practical device. That said, the technology will improve and one day, it may well be possible to print a useful and good looking clock on a domestic printer.

Frankly right now if Boeing wanted to, they could likely print off the engines to a 747 or whatever as single print jobs

Classic example of the kind of thing you don't want to print, because material properties are critical. Current metal printers can produce some very impressive looking parts, but they always have internal stresses and flaws that make them unsuitable for, for instance, jet turbines. Also, you do need to post process them.

[Mazon Del]

I have to disagree with you here.

3D printing does not give the same quality as other techniques, and never will. The resolution you get is worse than the resolution used to make the injection head, so the technique used to build the printer will always be better than the printer.

And 3d printed metal is basically crap. Making a metal piece is not as simple as it sounds, you have to think of grain size, thermal stress, surface treatment, and a million other things. 3D printed metal is either metal powder held together by glue, or sintered which means it's porous, fragile, has a huge surface area, and is unsuitable for many surface treatments.

And by the way, Boeing doesn't build engines, they buy them, as does Airbus and other airplane companies. It is a very different job to make a plane and to make a turboreactor, one that GE or Rolls Royce has been studying for the past 60 years.

I'll concede on the point of Boeing buying rather than building the engines, though that does technically just transfer my argument over to the companies that build the engines.

As far as 3D printed metal, I have to disagree with your statements primarily because of several reasons. First off, just about all those issues are things that can be taken into account for what you are making either with current tech or near future tech, there is a LOT of money going in towards solving all of this. Second, if the resulting metal really was crap, then you wouldn't have NASA switching most of their rocket nozzle production over to 3D printed nozzles or SpaceX constructing the vast majority of their SuperDraco engines in single-print 3D printing processes. I'll concede that as a result of the geometry of possible printed parts there may be some issues with surface treatments (internal cavities, etc), but that the idea that all metal printed parts have this issue is just wrong. You can always do post printing processes on them to provide them with appropriate finishes. Sand them down, buff them, etc.

The idea that the resolution of a printer will always be worse than the device that made it is quite false. Yes, a lathe will produce a more round object than something printed off a stepper style printer, and even quite likely one produced by newer servo style control. But not all printing technologies are created equal, just as not all manufacturing methods are. Different applications will provide different results. A lathe is likely to be more precise than a drill press, but lasers are getting more and more precise with their applications of energy. In plenty of instances lasers are the go-to if you have the cash. Laser sintering/curing printing WILL overtake bulk production methods in precision and I would need to do some research, but I would bet you they already have in some applications. Here is why. With something like a lathe, regardless of how precise your positioning and control systems are, the process IS still a purely mechanical process. This introduces certain unpredictable forces and vibrations that WILL limit precision of the device. Those same grains in the metal that you bring up, slicing into a new one is altogether not too different from a forces perspective on the scale of a full lathe. But when you get down to small enough scales it is causing movement within the cutting edge. Ever so slight movement. Can we dampen this out? If we have a way to predict it, maybe. But the issue comes from the fact that unless your cutting surface is super small, then you are going to be unable to fully dampen out the forces coming from two different grains entering the cutting surface at the same time.

Mechanical cutting techniques by the very nature of how they actually work are completely unpredictable on a small enough scale, which seems a little counterintuitive, but has some good reasons. [Apologies, this part may seem a little patronizing, but the background is important.] The question of if something can cut something else comes down to a few questions. How hard is each material? How much force is being applied from object A to object B? And what area is that force spread across? Depending on how deep you want to get, there are more variables, but these are the heart of the matter. How much force is very important simply because without enough of it, neither material is going to deform the other. What area is the force spread across is important because the actually important variable is force per square unit of size. IE: Pounds per square inch (PSI), or for the metrically inclined, grams per square millimeter. Fun fact, a 100 lbs woman in stiletto heels exerts more pressure on the ground at the point of contact than a 20+ ton Abrams tank does. This is because her weight (and thus the force of her standing there) is focused onto a tiny point, whereas tank treads spread the load around. Now that we know the force per square unit size, we look into the hardness of the objects. Whichever object is experiencing a PSI greater than its hardness allows it to resist, it will deform. Both objects may deform at the same time. So, if you made a blade that is as sharp as possible (one molecule wide blade), you have maximized the pressure you can exert with a given force because the size is as small as it can get. But remember, equal and opposites apply. The force acting on your cutting blade is the same that is being applied to the object to cut. Chances are good that you are not using some magical material with infinite hardness. As a result, you will experience a deformation of both the cutter (object A) and the object to cut (object . The blade will continue to deform (flatten) until it has reached an exposed surface area such that the PSI between objects A and B is no longer sufficient to deform object A. Below a certain scale the realities of molecules and how materials hold together and/or separate come into play against you. The surface of the cutter will not be flat perfectly, because this would mean that material pressing past the edges of the flat plane would again be exposed to near zero surface area and thus near infinite forces. Note, this 'rounding' is still on a super-fine scale so the blade is still quite sharp on the macro scale of your finger! So the result is that the cutter WILL assume a mostly rounded shape with an extremely small area of near perfect flatness. To tie this back into rockets (this being the KSP forums and all) we know that a flat object undergoing reentry will accumulate a layer of gas in front of it that simply cannot get out of the way in time for the next bit of matter. This is exactly the same as for cutting surfaces that have become rounded! As your blade cuts something fascinating occurs. I sadly was unable to find the videos that MIT made decades ago showing what I am about to describe, but I will do my best to paint a picture for you. As you build up this layer of material as your cutting blade 'undergoes re-entry through the other material' (sorry, couldn't resist that analogy) a curious thing happens with the buildup of material. The further away you get from the cutting tip, the thinner this buildup gets. Meaning that after a certain distance, the buildup has again formed itself into an atomically perfect blade! Right now you might be thinking that this is me proving Idobox's point since the cutting tool is cutting with atomically perfect blades. Except remember when I brought up those very same grains that Idobox used against me? They are very important here! Those same grains result in uneven forces being applied to this atomically perfect bow wave that the cutting blade has. This results in the bow wave collapsing almost in its entirety. So where does it go? Either into the surface of the part you are cutting (and thus ruining its perfect cut), into the ribbon or chips of metal coming off to the side, or more likely, both at the same time! It is entirely unpredictable! This establishes that in general, for a given hardness of your cutting tool, and a given hardness of your material to be cut, there is a minimum resolution beyond which you CANNOT reliably cut. This is the limit imposed on cutting tools. Unless we come up with some near infinitely hard material or nigh-2D energy blade, this is where the ability of mechanical machining processes to get more precise comes to an end. Period.

Now, this minimum cutting precision that I have mentioned is STILL on an absurdly fine scale. You will likely be unable to tell with your fingers that this is anything but smooth. But we have devices that can very easily detect just how unpredictably jagged even the smoothest of machined surfaces actually is. Fun fact. Don't try using these machines on your fingers. They are sharp. Very sharp. Now, how is it that 3D printing can overcome this limitation on physical processes? This is because 3D printing is not actually limited to the physical realm for all of its processes! Why is that? For technologies used in extrusion printers, they are entirely mechanical in nature. For printers utilizing laser sintering or curing, all of the hard work of precision is achieved almost entirely by electronic processes! For a great deal of laser technologies, they are steered via a galvo. A galvo is basically a mirror on an electric motor. Because of the aforementioned physical process limitations, the galvo can really only steer a beam of energy as precisely as the motor can turn. But this is not the only way to steer a beam. There exist technologies that allow for purely electronic steering of beams. Generally speaking, phased array systems are an example of such. There are no moving physical parts involved. Suddenly your precision becomes purely a matter of how perfect your crystals are (we are getting better and better at making them, and due to the very nature of how crystals work, it is possible to make a perfect crystal and KEEP it atomically pure) and how precisely you can control your electronics (IE, how fast your clocks are and how much noise you can dampen out of your power systems). So, what this means is that we can steer a beam with nigh-atomic precision in controlled circumstances (which a production area designed for this method would be). The new mechanical limit of the system is on your ability to lay down a thin layer of material, and to do so evenly. This is an area that any company researching non-extrusion printing technologies is investing most of their R&D budget into. Laying a single atom/molecule layer down perfectly evenly isn't easy to do, but it is at least possible to do within the tolerances of a couple atoms/molecules, unlike mechanical cutting on the level of atoms which simply cannot be done reliably.

TLDR: Mechanical processes have a limit that is basically impossible to overcome short of a nigh-infinitely hard material to use as a cutter. Printing processes can be almost entirely electronic in nature (except for the material being printed) and thus can achieve a precision far more precise than mechanical can.

(Edit: I always find it funny when I type up something like this in the 'Quick Post' window.)

Edited November 21, 2014 by Mazon Del
Amusement

[Idobox]

NASA doesn't print rocket motors, they printed a injector. 3D printing has some advantages, in that it allows very complex geometries (that can be lighter) and working with some metals that are otherwise very difficult to work with, because of either hardness, brittleness or high point of fusion. 3D printing is a useful tool, but not a magic bullet.

You posted a wall of text on the precision on cutting, but you forget the existence of forging, moulding, electrochemical and electro-thermal processes. Polishing can be result in surfaces that are so smooth the roughness is very hard to measure (telescope mirrors are polished mechanically, because it's still the best we can do).

If you take a piece of steel, heat it and then cool it, depending on the temperature curve you followed, you will get vastly different material properties, the size of the grains will differ, the intergrain properties too, the austensitic/martensitic phases... If you use a fast temperature changes, the properties will not be the same everywhere and you will get internal stresses. You can also direct the propagation of alloying metals and impurities through the metal.

With laser partial melt, you get something more cohesive than sintering, but you still end up with welded metal powder, which is very different from cast or machined metal.

In some cases it won't matter (let's say for figurines), for some casting metal is just too difficult/expensive (tungsten, some superalloys), for some the lighter, more complex geometries allowed by 3d printing overweight the limits of the material properties, and for some, 3d printing will be the only way (heterogeneous powder, metamaterials). But 3D printing metal will never give you the same properties as other techniques. Some stuff will be cast, some stuff will be machined, some stuff will electroformed, some stuff will be forged and some stuff will be 3D printed.

By the way, Russians engines are so much better than US ones because they have better metallurgy, allowing for LOx rich mixture to be used in turbopumps, which has all sorts of advantages.

On another side note, injection head for inkjet printers are often build by electro-forming on graphite shapes. The same technology is used with metal on silicon to make nano scale devices. Sure the process involves photolitography, but it's not 3D printing, and it really touches the absolute limits in terms of precision, with layer thickness that is not even a whole number of atoms.