Tuesday, September 15, 2015

Early lunar mining

 Returning to the theme of bootstrapping for a bit, let's examine what kind of material processing could be done with a modest payload. I'll cover two scenarios, lunar surface and captured asteroid; the first post will discuss the moon.

 As with all unproven technology, mass estimates for mining equipment are wild guesses. I'll be using the wild guesses of people smarter and/or better informed than myself. Most of the concepts presented are well-known; I've simply combined them a different way and extrapolated the results.

For those not interested in reading the wall of text to follow, here are my results:

3x haulers based on NASA chariot / lunar electric rover (1.2 ton each)
3x prospecting package: gamma spectrometer, neutron spectrometer, UV/VIS/IR spectrometer, magnetometer, robotic scoop (included in hauler mass)
2x excavator equipment package: bucket and cable rig, sized to fit hauler chassis (up to 3 tons each)
60kW power center (1 ton)
6x 90m² solar reflecting ovens with electrodes (0.5 tons each)
ore separation / benefication (0.5t)
cryogenic oxygen plant (4t)
tank press (0.5t)
radiators (2t)
electrical cables, 6ga/10.5mm, 4-conductor, 108v 3phase, 6kW, 8km (4t)
4000x cryogenic tank valves (0.2kg each)

total: 25.9t
(1 SLS or 2-3 Falcon heavy)

 - 2,400 tons of material processed per year, 10 tons per lit day
 - 840 1m³ filled oxygen tanks per year (958t LOX + 95.8t aluminum)
 - 128 1m³ filled water-ice tanks per year (120t H2O + 1.2t aluminum)
 - 527t metals (iron, additional aluminum, titanium, calcium) per year
 - 480t silicon
 - Unknown quantities N2, trace metals, other volatiles
 - ~217t refractory slag (radiation shielding); can be pressed, sintered and metal-wrapped

The supply of valves is enough for four years; they are brought along because they are too complex and risky to assemble in place. A pressed and sintered (or SLS + heat treated) pressure vessel is simple enough to be produced on-site even if welding is required. Sufficient oxygen will be available for burst testing the results before use.
If significant amounts of nitrogen are found, that would also be liquefied and tanked. Hydrogen would be reacted with oxygen and stored as water.
Cables can be eliminated at the cost of more batteries and shorter range.
Minimal additional equipment (1-2t) can enable production of thin-film solar panels with quartz front-glass and aluminum, titanium or iron backplates.

Back to the long form...

A moon expedition would produce huge amounts of oxygen. The main use for this would be to transport it to low-Earth orbit and pair it with hydrogen launched from Earth (or collected elsewhere) for fuel or for water. 1 ton of water contains 888.9kg oxygen and 111.1kg hydrogen; 1t of hydrogen from earth could be paired with 9t of oxygen from the Moon to yield 10 tons of fuel or water; this 10:1 ratio provides significant savings. For manned exploration of the moon, having oxygen available at each step of the trip saves much more than 9:1 since it does not have to be lifted from Earth and then sent to the Moon. For a cargo trip to Mars, the Earth departure stage calls for anywhere from 90 to 115 tons of fuel; using Lunar oxygen frees up 81-103 tons of payload to LEO or eliminates an entire SLS launch of fuel. A Mars campaign that requires four cargo trips and two manned trips could save six fuel launches or about $4.5 billion over six years. A single SLS launch could deliver a payload to the Moon capable of producing the ~600 tons of oxygen in one year as well as the vehicles necessary to deliver it to LEO; even if that mission costs $750 million for the rocket and another $1 billion for the payload it would save $2.75 billion while validating ISRU technology prior to its use on Mars.

 For a variety of reasons, the most useful place for a single mining operation on the moon is at either pole.

 Problem: The moon rotates at the same speed that it orbits Earth (it is tide-locked; the same face of the moon always faces Earth). Since the moon's orbit is about a month long, each day has two weeks of daylight and two weeks of night at the equator. That means enough power to run for two weeks of darkness has to be stored during the lit hours; most industrial processes would have to be halted.

 Solution: High peaks or crater rims near the poles can be sunlit for 80% of the time or more. (due to the axial tilt and geography there are no permanently-lit areas at ground level.)  A base at one of these locations could operate at full power for around 22 out of every 27 days.

 Problem: The moon's soil or regolith is bone dry to at least 30cm deep, and it seems unlikely that there is any bulk ice under most of the surface. Metals (titanium, aluminum, calcium, iron, magnesium), silicon and oxygen are abundant but hydrogen, carbon and nitrogen are exceedingly rare.

 Solution: There are craters whose bottoms are never lit by the sun, conveniently located next to the high-sunlight peaks. These areas get so cold that they can freeze water and volatile gases (including nitrogen); this is why they are called cold traps. There is strong evidence of water ice, hundreds of millions of tons of it, at the north polar craters and no reason to suspect it is not also abundant at the south polar craters (Shackleton crater).

The mission should target either the Shackleton crater rim near the south pole or the rim of Peary crater near the north pole.

 Remaining problems: many.

 Rich sites to be mined need to be identified with some kind of sensing instrument (or more likely a set of 2-3). Nearly all of the major minerals on the moon can be processed for oxygen, but some of their metals are more useful than others.
 Estimates for volatile concentrations are as high as 5% for the target areas. Let's assume the bulk excavation of material yields 1% volatiles and further assume that it is nearly all water (either free as ice or bound as hydrates). There is some free hydrogen from the solar wind and potentially traces of hydrocarbons and nitrogen-bearing organics, but without enough data to even guess at a concentration. It is possible that there is bulk water ice available; if so, it can simply be melted, filtered and frozen again in tanks with minimal energy requirements.
 Just like on Earth, craters on the Moon were caused by asteroid or comet impacts. Many of these impactors are rich in metals (iron-nickel-cobalt and platinum group metals), while others are rich in carbon. It is possible that these sites will serve as rich orebodies for these resources. There are also a few areas where mantle material is exposed; this boundary layer is rich in incompatible elements like rare earths.

 Products of this early mission would be water, hydrogen, oxygen, base metals, possibly ceramics and possibly semiconductors. Later work would advance to construction of integrated circuits, LEDs and multijunction solar cells in addition to worked metal (bar, sheet, tube stock).

 A mobile excavator would be required, probably a simple dragline. I've seen proposals that assumed the regolith would be powdery and easy to scrape up, but the Apollo reports indicate that under a surface dust layer the soil was surprisingly tough. Some think that the grains have settled due to vibration from impacts over the past few billion years. Regardless, the tough soil and the possibility of crystalline ice means the excavator needs to be able to break up soils. In the low gravity of the moon, this requires either a very large amount of mass to produce the necessary force or a novel solution of some kind; either way will require a lot of power. I'm fond of the idea of a rotary flail since the impact force can be closely controlled, but that introduces a mechanical part that will wear over time. I am also fond of the idea of using a sheet or bag to cover the active area, something to capture flying dust and debris to avoid messing with the atmosphere and surrounding areas. Presumably this would use electrostatics to capture dust.  Due to the high power requirements I assume this excavator will be connected to the power center with a cable up to a few km long; this is one of the heaviest single items on the list but it is less risky than batteries or beamed power.

 See NASA DRM5 numbers; they propose an excavator system that is under 1t all-inclusive and redundant, including onboard power via electrolysis.

 Once the raw material is excavated it has to be processed. Options include baking out the water at the excavator or hauling it back to a refinery. I prefer hauling it back to reduce thermal pollution (that is, to avoid cooking off the useful volatiles in the mining area). So, this introduces the need for a hauling rover of some kind. The output of the excavator will drop directly into the bed of the hauler, with an electrostatic dust sheet over the top.

 The hauler will deliver regolith to the processing center. Samples will be analyzed for content in order to determine the most efficient processing plan. Volatiles like water will be baked out using solar concentrators at 600-1000 °C and then cryogenically separated. The remaining ore is now stable and can be stockpiled for later processing if needed. Iron-rich nodules and fragments will be removed with a magnetic rake. The remaining grains can be sorted with a rotary device into bins based on grain strength (aluminum-rich, titanium-rich, glasses) anywhere from 50% to 90% enriched.

 Many proposals assume the use of a reactant gas (usually hydrogen) which will be recycled; this gas is used to extract oxygen at lower temperatures than would be required otherwise. This seems unreasonably optimistic; hydrogen gas is difficult to contain on Earth, but we're talking about a hard vacuum environment with the most hostile abrasives we've ever encountered in the natural world. It will have to handle hundreds to thousands of open/close cycles with a perfect seal in the face of 1500+ °C temperature swings. Add the lack of maintenance or human operators for adjustments plus the need for vacuum pumps with both high volume and very low pressure and that all adds up to heavy, complex and unreliable.
 I choose to assume a brute force method of direct molten electrolysis. Instead of using hydrogen and reducing iron oxides at 1600 °C, the processor will melt the ore grains at 2200-2500 ­°C and electrolyze the melt with tungsten electrodes. This will produce oxygen gas at one end of the cell and reduced metal at the other end. For the lighter metals (aluminum, calcium, sodium) the cell temperature is high enough to vaporize them; the metal vapor can be collected on cold plates and scraped off as fine grains or it can be used directly for vapor deposition. Heavier metals (iron, titanium) will accumulate as melt and can be tapped and formed into bars. High-temp refractory ceramics like magnesia are out of reach of this method but might be available to a solid-phase electrolysis process. The oxygen is passed back through the incoming ore stream for heat exchange; any unreacted contaminants will definitely be oxidized in the process.
 Concentrated ores can be heat-treated further as a form of distillation to remove materials in order of their melting points; a carefully-designed feed mechanism can pass the ore through one oven, allow the lower-temp material to melt, then deliver the higher-temp material to another oven. The temperature of the refining furnace can also be adjusted to evaporate out metals in sequence with reasonably high purity. One additional control is the voltage of the electrolysis cell; if necessary the oven can be operated as a batch process using appropriate voltages to reduce each metal in sequence.
 The processor requires a very high temperature furnace. This will be made of magnesia (magnesium oxide, melting point 2852°C) and heated with a mix of joule heating and concentrated sunlight, which will require a large reflecting area of aluminized mylar. The magnesia is very heavy; it may be possible to construct a suitable trough on-site from regolith baked, pressed and sintered into a useful form. At this stage in the process the ore grains have been heated to recover most of the volatile content; the melt can be performed in open vacuum for direct solar heating with only the electrode sections enclosed to capture products. This avoids the need for an absurdly high-temperature transparent material. The furnace will be continuous process; accumulated solids are scooped out and dumped (and possibly used as a heat source for the first baking step). These can be compressed in the tank press to form radiation shielding blocks if desired. Oxygen is continuously produced and passed back up the ore stream, then cooled and passed to the cryogenic plant. Reduced metal either vaporizes and is condensed on a cold plate or deposition target or is drawn off as a liquid from the bottom of the melt and cast into the desired form.

 As a specific example: aluminum-rich grains will be melted (2072°C) in a magnesia crucible by concentrated sunlight and then electrolyzed (with tungsten, tungsten carbide or molybdenum electrodes) into oxygen and aluminum metal. The temperature can be further increased to 2470°C to produce aluminum vapor; thin-film conductive/reflective coatings or vapor deposition can be performed directly in the furnace. Aluminum metal parts can be controllably oxidized using the hot oxygen stream to form alumina surface coatings for abrasion resistance. This requires about 3.2MJ/kg from 0°C to 2100°C; 80-90% of this can be from sunlight, about 40m² of reflectors and 6kW for 1 kg per minute capacity. 52.9% of the yield is aluminum and 47.1% is oxygen by mass, barring any other metal oxides in the melt.

 As it turns out, tungsten is not suitable for this process over the long term because it is oxidized. An inert surface layer of iridium is required on the anode (oxygen-generating), while molybdenum is sufficient for the cathode (metal-generating). The anode layer needs to be thick enough to resist abrasion from any unmelted grains. A cylinder furnace of magnesia with molybdenum cathode strips along the bottom and an iridium-coated tungsten anode bar along the long axis will be used; the entire furnace is sealed and heated by concentrated sunlight from the outside. Preheated ore grains are loaded at the top of the 'near' end with an augur. A baffle keeps gases contained but the melt's surface is exposed to vacuum at this end; if this volume is enclosed then a cold plate could capture volatiles that were missed by other process steps. Oxygen is tapped at the top of the far end, while molten metal is tapped near the bottom of the far end. The endcaps of the furnace can be removed so any accumulated slag or debris can be removed and electrodes can be replaced.
 In order to boil the aluminum metal, the entire furnace's operating temperature is driven higher. Care must be taken to keep the oxygen gas and aluminum gas separate; if this mode of operation is desired then the internal structure should be shaped like a U, with cathode and anode in separate arms.
 There are ongoing efforts within NASA and within the steel industry for developing robust molten oxide electrolysis cells; a prototype molten regolith cell might contain 0.02m³ of volume and consume 3-5 kW of electricity as a 100% joule-heated furnace (with no mention of cycle time). My proposal needs to process 10m³ per day but can obtain very large amounts of heat externally; as a result my design can place the electrodes closer together, reducing the resistance of the melt and thus the electrical losses to heat.

This method can be used for extracting iron, calcium, sodium and titanium from their oxides as well. While useless on earth, pure calcium is an excellent conductor in dry, oxygen-free environments and could substitute for aluminum in PV conductor lines. Sodium and calcium are potential scavengers for corrosive volatiles like chlorine and fluorine, but their vigorous reaction with water calls for careful process management.

 Silica grains can be used to make clear glass. Thin quartz glass sheets are a starting point for making PV cells (the surface protective layer); conductive lines of aluminum are vapor-deposited onto the glass, then doped silicon, then a thicker layer of aluminum. These are not the most efficient devices (~9%), but they are very simple and straightforward to manufacture. The silicon can be extracted from silica just like the aluminum is refined, but at a lower temperature (1713°C).

 With a goal of 1 ton of 1% icy regolith processed per day we should be producing 10kg water, 400kg oxygen, 200kg silicon, 120kg iron, 60kg aluminum and about 260kg of other metals and unprocessable byproducts. Specific ratios of silicon, iron, aluminum and titanium depend on what minerals are present at the excavation site, but it is very difficult to predict in advance what the ratios will be at a specific site. This processing will require roughly 30m² of solar reflector area and 4.5kW of electrical power. If we assume the initial power system alpha is 200 watts per kg (since it needs to withstand lunar gravity), that's only 22.5kg of solar panels.
 Pushing things a bit farther, 10 tons of ore per day would require 45kW of power (225kg) and 400m² of reflector area (about 8kg plus supports). A minimum work cycle of 20 days per sol means about 2 tons of water, 80 tons of oxygen and 52 tons of metals per sol (27 days). If the excavator manages to produce ore with 5% ice then the yield is 10 tons of water per sol; if bulk ice is discovered then a different process will be needed. It would be possible to process larger quantities of ore for volatiles and only process a portion of the desired metals; extracting water takes a tiny fraction of the energy needed to melt and electrolyze metal oxides.

See the NASA DRM5 numbers for ISRU equipment; to produce 56kg H2O per day from 3% water-content soil required 413kg of equipment and 2.02 kW of power. That's roughly 1.8 tons ore per day; multiply by six to hit the 10-ton target and that's 2.5t of equipment and 12kW of power for the entire process end to end. Most of the power is used for process heat, so substantial savings can still be obtained with solar reflectors. This would produce a supply of dry granular material for later processing or for direct use as bulk shielding.

 The excavator needs to remove a bit under 7kg per minute. Bulk density is around 2g/cc (2ton/m³). A 100-liter bucket would have to be filled about every 28 minutes. If the hauler holds 1 ton (cube 80cm per side) then a full round trip (load, deliver, unload, return) must take 2 hours and 24 minutes or less. Assuming load and unload take 12 minutes that leaves 1 hour for the drive one-way; better yet, let's assign 1.5 hours for the loaded trip and 30 minutes for the empty trip. At 5 m/s maximum empty speed (1.67m/s loaded speed) that gives a range of 9km between refinery and excavator.
 If the hauler holds three tons (100x100x150cm) then a full trip is at most 7 hours 12 minutes. Using the same 12 minutes to load or unload and 3/4 of the trip time loaded, that's 5 hours 6 minutes for the loaded trip and 1 hour 42 minutes for the return trip. Using the same speeds (5m/s empty, 1.67m/s full) that gives a range of 30.6km.
 An alternative is to use both haulers in rotation; one is stationed at the excavator getting filled while the other travels to the refinery and back. There would be enough time for prospecting during the return trip.

NASA already has an electric rover design in this size range; 1 ton of vehicle mass for 3 tons of payload. That design's maximum speed is quoted several different ways, but at least one reference uses 15mph (6.7m/s) as high-gear speed, presumably for just the chassis, and several others use 10km/h (2.78m/s) as the top speed while loaded with a mobile habitat module. I would add prospecting sensors, a small scoop and a blade and auger.

For reasons of efficiency, the excavator should use a hauler chassis with an excavation package mounted on top. A spare chassis can be shipped in case either the hauler or the excavator fails. Resupply missions can carry modular parts. Two excavation packages will be shipped with the initial mission, both using a drag line arrangement that minimizes repositioning time. (A set of two pylons with pulleys, so the bucket can be moved anywhere within a triangle defined by the two pylons and the hauler base. A 50m triangle with 450m of cable can cover 1082m² of territory; for excavation to 2m that's 2164m³ or about 4300 tons, more than a year of production.) All three haulers will be equipped with a blade and augur. In fact, the initial phase would involve all three vehicles using just their blades to clear a landing site for cargo spacecraft, followed by a short roving mission to identify high-ice sites and then leveling a basic road to the best site. This will provide some ground truth on solar wind ions and other volatiles in the upper regolith plus validate several approaches to lunar base site clearing for manned missions.

Harvested metal (primarily aluminum) is pressed into cryogenic tanks using multipurpose valves brought from Earth (solenoid with integrated pressure relief). These tanks are pressure-tested and then filled with liquefied oxygen. Filled tanks are stored in a permanently-shadowed crater for zero boiloff. Tanks are shipped to L1 for about 2.5km/s dV in a dedicated lander, then to LEO for a further 0.8km/s in an orbital tug with heat shield or 3.8km/s propulsive-only. Ideally a launch sling would be used to deliver the tanks to L1/L2 and avoid the need to burn propellant.

 Once this architecture is validated it could be reapplied at other sites and for other purposes. The haulers could be shipped to a future human lunar base to clear landing sites, roads and construction sites, plus accumulate material for radiation shielding and later bury the habitats. The processing equipment could manufacture solar panels, heat radiators, oxygen, structural materials, etc. in advance of crew arrival so the mass to be shipped from Earth can be minimized.
 Similar equipment could be shipped to Mars for the same purpose; the initial cargo mission could deliver equipment for site preparation and atmosphere processing. Mars has a much more complete chemical environment, so a much more useful selection of products can be produced.

Again, NASA DRM5 has useful numbers for PV power systems. A single PV/RFC (regenerative fuel cell) module was 4.5 tons and provided 5kW of fuel cell power and 290m² of collector area at 29% efficiency. Under lunar conditions (no atmosphere, 1366W/m², ground return ignored) that array produces 396 kW gross, roughly 336kW of conditioned power. The whole system could be reduced to one fifth, massing 0.9t, providing 1kW of fuel cell power and 67.2kW of PV power. This would be abundant power for daytime operations but probably insufficient night-time power. However, the fuel cell's peak output would be more than sufficient, so extending operation to several days would require only larger tanks for water, O2 and H2. This is reasonably in line with my original estimate of 1t for 60kW.

Further, the DRM5 offers a nuclear alternative. A 30kWe nuclear reactor masses 7.8t and would enable continuous operation. I think this would be an enabling technology that would completely change the architecture of a lunar mining operation. Night on Luna is the enemy for solar-powered architectures, leading to deep thermal cycling and heavy battery or fuel cell requirements. It also greatly expands the potential locations for a base. There would still be a need for rover excavators; nuclear power means a lot of radiated heat that we want to keep away from the cold-trap volatiles we are most interested in harvesting. One benefit is that the reactor's waste heat can be used as a first stage heat source for ore processing. Using JIMO (project Promethius) data, the proposed brayton cycle reactor has an end to end efficiency of 18.35% from thermal to conditioned electrical power at the PMAD output. The Mars design has slightly improved efficiencies, but JIMO performance would be sufficient. Let's assume the difference means a 30kWe reactor would mass 8t even and spec 163.5kWt. This reactor's coolant is 920-950K at the recuperator, a very convenient source of process heat at 650-670 °C and around 133kWt available.
 Lunar soil's specific heat is 0.88 kJ/kg*K at 350K; data is sparse for higher temperatures but can be expected to increase somewhat up to 950K; let's call it double or about 1.7kJ/kg*K. That would mean the available heat could cook soil from 90K to 950K at a rate of 5.46kg per minute. 10 tons processed per day would be 6.94kg per minute, so we are in the right ballpark. A counterflow heat exchanger running between the incoming ore and outgoing ore would only need to recover about 22% of the heat to close the thermal requirements.
 If relatively pure water ice is available, it can be taken from 90K to 275K (2 °C) for ~334kJ/kg + 334kJ/kg to melt it. The reactor could process 23.9kg of water ice per minute; this application would use heat at the radiator return (~390K) rather than the recuperator for more efficient operation. This approach could handle up to 34.4 tons of water ice a day.
 Both of these modes of operation would leave the entire 30kW of electrical power available for other purposes

No comments:

Post a Comment