· · · · · · · ◦ ◦ ◦ · · · · · · ·
The Rising Sea is building massive self-replicating double-column gantry mills: economically useful machines that can reproduce.
Our aim is to in-house the manufacturing of all materials, feedstocks, and technologies necessary for human civilization, from oxides fresh-out-of-the-ground to orbital launch vehicles.
We are beginning with precision machine tools because primary steel, of which we produce globally about 1.8 Gt/y, is perhaps the single fundamental pillar upon which our civilization stands, along with cement (4 Gt/y), plastics (400 Mt/y), and ammonia (200 Mt/y), and in order to do anything useful with steel, you need precision.
We are not a technology company, we are a science company.
Our ultimate aim is to accelerate fundamental research in the basic sciences to improve the standard of living.
We believe that dramatically reducing the iteration time needed for physical prototyping is the best way to achieve this, since almost all scientific progress is bottlenecked on its fiscal, material, energetic, and temporal costs.
And because you must iterate on your production infrastructure to iterate on your product, self-asssembly is the clearest approach to minimizing cycle time, since the production infrastructure is the product.
An NC (numerical control) lathe is a machine for making radially symmetric parts. It spins a piece of (usually metal) barstock as a computer-controlled cutting tool carves a path along which material is removed. Although it may seem like a specialized tool, lathes are extraordinarily critical for human civilization because they enable us to construct the prime movers which power everything: internal combustion engines, AC-polyphase induction motors, steam turbogenerators, combined-cycle gas turbines, and more generally, all engines with spindles or axles. The high rpm needed for these devices to function would be impossible without precision radially symmetric machining of steel and high-temperature alloys.
Lathes are one of three fundamental machine tool variants from which all precision in manmade artifacts is derived. A mill, which is what people usually imagine when they think of the phrase, "5-axis CNC machine", performs in some sense the inverse operation of the lathe. While a lathe cuts a rotating workpiece with a stationary tool, a mill cuts a stationary workpiece with a rotating tool.
The third machine tool variant is a grinder: a machine which basically rubs the workpiece with a rock (stone grinding wheel) to achieve a surface finish that is impossible by milling or turning alone. The surface finish on a part has importance far beyond aesthetics. It is only through this process that e.g. optics for laser interferometers can be produced, which are mankind's most precise distance-measuring tools and the key components in, for example, the LIGO facility for detecting gravitational waves and ASML/Zeiss's mechanism for positioning silicon wafers within sub-nanometer tolerances for EUV exposure.
These three machine tools (the mill, grinder, and lathe) are often referred to as "mother machines", since they are needed to produce nearly every other machine or product, and because it is nearly impossible to produce any one of them without already having use of all three. They can in some sense be considered fixed points of manufacturing. (In mathematics, a fixed point of an endomorphism [function whose domain and codomain are the same space] f is some x such that f(x) = x.)
To give some examples of their importance, consider three common products not normally associated with machining: a t-shirt, a toaster, and an America wood-frame 2-storey house. The t-shirt is not made of metal (obviously), but the tri-blend polyester-cotton-rayon fabric it's made of is produced on a steel loom whose critical moving parts are machined on a mill or lathe. It is cut and sewn using precision steel devices and the raw cotton fiber is harvested using steel machinery and grown using urea synthetic nitrogenous fertilizers produced via the Haber-Bosch process in pressure vessels inside a massive industrial facility full of machined steel components. The synthetic fabrics are produced via polymerization from petrochemical feedstocks produced via cracking or steam reforming of hydrocarbons distilled in massive fractional distillation columns. All of this infrastructure requires precision steel components.
The toaster's plastic housing is produced in a massive steel injection-molding machine using >100t of pressure from a mold that was, of course, precision milled from steel. We could continue ad nauseam about its innards.
As for the house, Vaclav Smil writes:
"...even if a building were to be constructed without any steel members or without any nails, by using doweled wood components, the wood would have to be cut by steel saws, and the building’s basement would have to be dug by steel machinery, or at least by steel shovels."
And of course those saws, machinery, and shovels cannot be manufactured economically without mills and lathes.
So back to what we're actually building. A mill-turn (i.e. multitasking) machine is one that can do both milling and turning (what a lathe does) within the same work envelope. Inclusion of grinding functionality combines all three functions into one device, which is rare but not uncommon. (And very expensive! Multitasking machines run from $400,000 up into the millions.) However, just because a machine can manage all three operations does not mean it is some kind of "universal fabricator." Differences in achievable tolerances, geometric degrees-of-freedom, and above all maximum workpiece sizes mean that no extant device on the market can perform the machining needed for all of its parts. However, mill-turns are generally considered the most capable machines for workpiece volumes measured in single-digit cubic meters or smaller.
Some of the most well-known manufacturers are Okuma, Yamazaki-Mazak, DMG Mori, Daewoo, and Hwacheon. There are more precise machines, like those made by Kern Microtechnik in Switzerland or Mitsui Seiki (whose jig grinders finished the mirrors for the James Webb Space Telescope) in Japan, but these machines are more specialized and thus limited in flexibility of function. The larger-volume machine tools in the multitasking category have masses around 25,000-40,000 kg, the vast majority of which is accounted for by a grey (an adjective which describes a specific crystalline structure) cast iron base upon which linear guides (ball-bearing rails) are mounted which facilitate low-friction sliding of heavy moving parts.
Anyone who suggests that it is straightforward to make these machines lighter without crippling capability is very likely talking nonsense. Their rigidity and thermal expansion geometry is central to their basic function and after nearly a century of research by manufacturers, a large mass of cast iron is still the most cost effective way to achieve this. This should be unsurprising, since cast iron is remarkably cheap at only $100-150/mt and has excellent vibration-damping properties which are absent in fabricated steel beds. To remove material at economically viable speeds, the massive turret/carriage/gantry/spindle (depending on the machine type) must move very quickly and rotate at very high rpm. It is a high-school-level physics problem to compute the Young's modulus needed to hold sub-10-micron tolerance under these conditions.
The necessity of constructing a massive grey cast iron machine tool bed means that any serious attempt at self-assembly must include the foundry equipment necessary to shape and compress the sand mold and cores, heat the charged iron to its melting point of ~1535 C, postprocess the raw casting, and machine extremely precise flats upon which to mount the linear guides. This last part presents a considerable difficulty because most machine tool beds are machined and ground on house-sized double-column gantry mills so large they require custom foundations sometimes anchored to bedrock. In order to make a machine that can mill its own base, it may be necessary to weld the base together from multiple parts, which can potentially cause thermal problems but is not uncommon for very large tools. In addition, most high-end machine tool bases have flats which are hand-scraped using an optical surface plate as a reference in order to avoid the inevitable thermal warping caused by grinding. Hand-scraping is exactly what it sounds like: a worker holds a baseball-bat-shaped tool with a piece of carbide stuck to the end between their legs and scrapes ultra-thin chips of metal off the surface, periodically checking flatness by applying color to a surface plate and rubbing it on the in-progress piece. As an aside, this produces a quite beautiful irregular tiled pattern that machinists have come to associate with extreme precision surfaces.
Optical surface plates can be considered the source of precision from which the machine tools draw. You might ask yourself how one can make a more precise machine with a less precise machine? How do you bootstrap precision? The answer is complicated but basically boils down to flat surfaces, optics, and mathematical feedback.
Suppose you have some rocks and you'd like to make a flat rock. You may try rubbing two rocks together for a long time, but you need only try this once to realize it does not preclude the possibility of ending up with a mated concave-convex surface: a radius, basically. However, add a third rock and alternate the surfaces (placing convex-to-convex, and so on), and you create a system of geometric feedback which results in arbitrarily flat surfaces, limited only by the "grain size" of the material being ground.
Beyond the spatial issue of the base, a water-cooled induction furnace must be constructed in order to melt the charge of iron, and refractory (heat-resistant, often ceramic) material must be sourced to line it. In addition to the hollow copper coil that wraps the crucible, large banks of capacitors must also be water cooled in standard designs. Both the furnace and the lathe will require high voltage (at least 240V but preferably 480V) 3-phase power supplying substantial current. Making these devices independent of the grid is wholly unnecessary since it is trivial (CCGTs yield $10/MWh "portable" electricity) and industrial grid electricity is reasonably cheap and readily available in all locations where these machines would be initially useful to run.
Beyond the base, mill-turn machines contain thousands of parts and subassemblies, notably 10-60hp servos, stepper motors, or linear motors, recirculating ballscrews (a ball-bearing device for translating rotational motion into linear motion) for positioning with zero backlash, hydraulic coolant systems, compressors, automatic tool changers and turrets, pallet changers, integrated coordinate measuring machines (CMMs), glass-scale linear encoders, a variety of gearing, multiple spindle assemblies (essentially powered axles that are remarkable rigid and concentric), myriad workholding devices, reservoirs for oils and lubricants, power transformers, servo amps, PCBs, programmable logic controllers (PLCs) for the servos, a mess of cabling, indictor lights, and a control unit with display and a wide variety of physical buttons.
And in some ways most critical are the (usually tungsten carbide) indexable cutting inserts or endmills and grinding wheels. An endmill is essentially a drill bit that's built for lateral cutting, and an insert is a small piece of sharp, hard metal mounted to a larger tool that can be replaced in lieu of making the entire tool geometry out of carbide.
Although they are the hardest manmade objects outside of synthetic diamonds, these sharp little things are called consumables: they have a finite lifetime and a machine needs a continuous supply of a dizzying variety of them in order to do useful work. In professional production environments, speeds and feeds are optimized for standard tool lifetimes as short as 50 minutes!
The inserts are often coated (via gas deposition) in a several nanometers of some even harder material. TiN (which has a gold-colored finish and is often used on cheap consumer-grade drill bits) or AlTiN are common. The melting point of tungsten is too high for economical casting, and so the tungsten carbide (WC) blanks are produced by sintering a mixture of tungsten powder, carbide powder, and a bit of cobalt binder (also, no surprise, in the form of powder). The process of producing sufficiently pure tungsten powder from raw oxide is very complicated and requires a large number of electrochemical purification steps, passing through intermediate forms, notably ammonium paratungstate (APT), each of which removes a common impurity. Perhaps the most well-known and highly regarded manufacturer is Sandvik-Coromant in Sweden, which makes a large proportion of its stock from recycled carbide.
Clearly, a first-pass at a useful self-replicating industrial prime mover will need to classify several critical parts as raw inputs initially. The tungsten carbide consumables are good candidates, as are integrated circuits and large swaths of the other cheaply-available electronic components. Even a cursory look at the economics of primary steelmaking rules out using raw ores/oxides as feedstocks: industrial-scale foundries like Alcoa, US steel, Nippon Steel and POSCO operate on such small margins that they've been dangerously close to or outright unprofitable in recent decades. Margins have moved closer to the mining operations.
The Tale of Thomas Newcomen
Long ago, in the age of iron, before mankind left forever to wander the stars, there lived an Englishman called Thomas Newcomen. As a young man his ambition was great, and he set out to steal a treasure from Heaven that would bestow prosperity on the people of the world. For years he searched far and wide, seeking wisdom in all places from remote uncharted villages to ancient buried libraries.
Eventually, at the top of an adamant tower, he came to a workshop whose door had been left ajar. Upon venturing inside, he found an array of complex and subtle machines.
Not wishing to be discovered, he took the first thing he laid eyes on from the workbench, slipped it into his pocket, and retreated back the way he had come.
When the Master of All Good Workmen discovered Newcomen's theft, his anger was fierce.
He laid a curse upon the people of Earth: that they'd reap sorrow and joy in equal measure from this brazen act.
And so Newcomen stole the first Heat Engine from God, and with it the seed of our modern economies of scale.
Mankind bought for itself the ability to transform energy at unprecendented scales, but at great cost: as scientific infrastructure became complex and capable, it also became costly and slow.
In exchange for a few centuries of prosperity, we were left with a fragmented world, where progress on the frontier of physics, chemistry, materials science, and engineering was restricted to small numbers of sprawling, expensive projects.
Newcomen's curse set our civilization back 1000 years. It may be hard to believe it, but during this period, the second dark age, it was actually cheaper to ship technology from the other side of the Earth than it was to compile it in your own backyard. The enlightenment that was later brought on by the democratization of physical scientific innovation came only when humanity began to see economies-of-scale for what they truly are: rigor mortis. The first early signs of death for a civilization.