SShielded Dumbbell L5 Settlement
Pekka Janhunen a,b,c, ∗ a Finnish Meteorological Institute, Helsinki, Finland b Also at Tartu Observatory, Tartu, Estonia c Also at Aurora Propulsion Technologies Oy, Espoo, Finland
Abstract
We present a two-sphere dumbbell configuration of a rotating settlement at Earth-Moon L5. The two-sphere con-figuration is chosen to minimize the radiation shielding mass which dominates the mass budget. The settlement hasmax 20 mSv / year radiation conditions and 1 g artificial gravity. If made for 200 people, it weighs 89000 tonnes andprovides 60 m of floorspace per person. The radiation shield is made of asteroid rock, augmented by a water layerwith 2 % of the mass for neutron moderation, and a thin boron-10 layer for capturing the thermalized neutrons. Weanalyze the propulsion options for moving the material from asteroids to L5. The FFC Cambridge process can beused to extract oxygen from asteroid regolith. The oxygen is then used as Electric Propulsion propellant. One canalso find a water-bearing asteroid and use water for the same purpose. If one wants to avoid propellant extraction, onecan use a fleet of electric sails. The settlers fund their project by producing and selling new settlements by zero-delayteleoperation in the nearby robotic factory which they own. The economic case looks promising if LEO launch costsdrop below ∼ $300 / kg. Keywords:
L5 space settlement, radiation shielding
1. Introduction
To live permanently in space, a human being needsair, food, radiation shielding, earthlike gravity, and asu ffi cient number of fellow settlers and living space.All requirements can be satisfied in rotating free-spacehabitats made of asteroid or lunar materials as proposedby Gerard O’Neill in his pioneering works [1, 2]. Suchunplanetary living is attractive because it o ff ers a way toavoid common natural hazards such as hurricanes, vol-canism, earthquakes and wildfires. It is also attractivefrom the longer term population and economic growthpoints of view. There is so much material in the aster-oid belt (2 . · kg) that if made into settlements, itallows several orders of magnitude growth in the humanpopulation.One could build an orbital settlement in equatoriallow Earth orbit (ELEO) with much lower mass thanelsewhere, because in ELEO the Earth’s magnetic fieldprotects rather well against cosmic rays and solar pro-tons [3]. However, in LEO there is the risk of orbital ∗ Corresponding author
Email address: [email protected] (Pekka Janhunen)
URL: (Pekka Janhunen) debris. For example, recently the insurance companyAssure Space stopped o ff ering collision risk coveragepolicies in LEO [4]. There is also the issue of having toperform a targeted reentry when done with the facility.Letting it fall freely would be a public safety issue forpeople living near the equator.To avoid these issues, in this paper we consider a set-tlement at the Earth-Moon Lagrange L5 (or L4) point.The L5 point o ff ers Apollo-like short traveltime fromEarth, so that the transfer vehicle does not need muchradiation shielding. Satellite orbits around ∼ ∼ We write “L5 point” for brevity. In reality the orbit wandersaround the L5 or L4 point with large amplitude [5].
NSS Space Settlement Journal April 7, 2020 a r X i v : . [ phy s i c s . pop - ph ] A p r ause radiation shielding dominates the mass, an opti-mal entry-level configuration is a dumbbell comprisingtwo spheres. We point out that only the radiation shield-ing mass needs to be sourced from asteroids in the firstphase. We analyze a number of propulsion options formoving the material and do costing analysis. We closethe paper by discussion and summary.
2. Radiation shielding
Galactic cosmic rays (GCRs) have higher energiesthan solar energetic particles. For good radiation pro-tection, the GCR flux must be significantly suppressed,and this requires several tonnes of mass per square me-ter. At such shielding thicknesses the solar energeticparticles are suppressed almost entirely so they can beignored in the analysis.Globus and Strout [6] used the OLTARIS tool to sim-ulate the GCR equivalent radiation doses (millisievertrates) behind various thicknesses of di ff erent materials(Globus and Strout [6], Table 2). They recommend 20mSv / year as the equivalent dose level during the solarminimum (of 2010) when the galactic radiation is atmaximum, a value which we also adopt here.Here we also use the OLTARIS tool. The tool sup-ports two geometries: a slab and a sphere. In the slabgeometry, the dose between two adjacent infinite platesis predicted, where each plate has the given shieldingthickness. In the sphere geometry, the program pre-dicts the dose at the center of a solid sphere whose ra-dius is equal to the given shielding thickness. The dosepredicted using the slab geometry is smaller (typicallyabout two times smaller) than that using the sphere ge-ometry, because in the slab geometry, many of the cos-mic rays enter obliquely to the shield and so move alonger distance within the shield. For a hollow sphere,in the limit where the inner radius is much larger thanthe shell thickness, the dose at the inner wall approachesasymptotically the slab geometry prediction. The doseat the centerpoint of a hollow sphere does not depend onthe sphere’s inner radius, so it can be calculated by as-suming zero inner radius. Thus for a hollow sphere, thedose at the center is larger than the dose near the walls.The reason is that center-reaching cosmic rays passthrough the shell perpendicularly regardless of their ar-rival direction, whereas near the wall typical rays mustpass through the shell obliquely, experiencing more at-tenuation.For radiation shields, Z-grading is in general useful,that is, using high-Z materials as the outer layer and pro-gressively lower Z materials as one goes inwards. Wewant the bulk of the shield to be asteroid regolith. We also need structural material, which we assume to besteel, which is iron to a good approximation. We putthe steel as the outermost layer since iron’s mean atomicmass is larger than that of regolith. Inside the regolithwe put a layer of water whose mass is 2 % of the massof the regolith and water combined. It is the intentionthat this amount of water can be obtained from the as-teroid regolith by heating it. If not, the water can bebrought from Earth. The dry regolith layer is modeledby OLTARIS’ lunar regolith option.Part of the equivalent dose consists of neutrons spal-lated from the regolith by cosmic rays. The water layermoderates these neutrons. As the innermost layer weadd a thin 1 kg / m layer of boron-10. (If one wants toavoid isotope separation, one can use 5 kg / m of nat-ural boron which is 20 % B-10 and 80 % B-11.) Thisisotope absorbs neutrons e ffi ciently, especially thermalneutrons. The water moderates the neutrons to be ab-sorbed by the boron.Table 1 defines the layers of our wall. This shield wasdesigned to limit radiation to 20 mSv / year equivalentdose at the center of the sphere during worst case, i.e.,solar minimum. The equivalent doses were computedfor “Female Adult Voxel” phantom [7]. Table 1: 20 mSv / year shield using 2 % of water. Material, thickness kg / m RoleIron, 2.3 cm 180 Structural wallDry regolith, 3.34 m 8683 ShieldWater, 17.7 cm 177 Neutron moderatorBoron-10, 0.4 mm 1 Neutron absorberTotal, 3.5 m 9041For the solar maximum (of 2001) conditions, theequivalent dose is 25 % smaller (14.87 mSv / year). Forsolar minimum, a slab geometry calculation shows thatnear the inner wall the equivalent dose is 50 % smaller(9.97 mSv / year) than at the center. During solar max-imum the inner wall equivalent dose drops to 7.53mSv / year.Globus and Strout [6] also require that the absorbeddose for pregnant women be less than 6.6 mGy / year, or5 mGy per pregnancy. In our case this condition is sat-isfied since the absorbed dose at the center of the sphereduring solar minimum is 4.0 mGy / year.
3. Mass-optimal geometry
A sphere has the minimal surface area per volumeand is thus the best geometry for minimizing shieldingmass. To include artificial gravity, we need two spheres2
NSS Space Settlement Journal otating about each other in a dumbbell configuration(Fig. 1a). We select a baseline rotation rate of 2 rpm(revolutions per minute), which is probably a conserva-tive choice regarding avoidance of motion sickness [9].With 1 g artificial gravity, 2 rpm corresponds to rotationradius of R =
230 m. The main structural element is thetruss that connects the spheres. It carries the centrifugalload of the heavy spheres. It also acts as the shaft ofan elevator that provides access from the living spheresto a central docking port. There are two docking portsfor redundancy, upward and downward in Fig. 1. Thedocking ports are essential because they are the way toenter and exit the settlement. To ease the docking of theconnecting spacecraft, the docking ports are located onthe axis of rotation.a)b)
Figure 1: (a) Dumbbell settlement with elevator shaft and centraldocking ports, (b) with ringroad, cylindrical solar panels and green-house areas indicated (green).
To move from one sphere to the other, one can use theelevators, but then one experiences temporary weight-lessness in the central region, which is an inconve-nience. To eliminate this problem and to provide re-dundancy in routing, we add a pressurized ring-shapedtube that hosts a road. The ringroad is at constant radial distance from the rotation axis so its user does not needto move uphill or downhill in artificial gravity.To produce food as part of the closed ecosystem, weadd artificially illuminated greenhouses (Fig. 1). Thegreenhouses rest on the cylinder on which solar pan-els are mounted. The greenhouses are served by theringroad and we place them 90 ◦ o ff the heavy livingspheres to make the azimuthal mass distribution moreuniform. We assume that food production needs 16 kWof electrical energy per person, so 3.2 MW total for pop-ulation of 200. This corresponds to 2500 kcal per dayper person of food plus 30 % margin, and 1 % e ffi ciencyin converting greenhouse electrical energy into edibleenergy of the crops. Most of the energy is dissipated in-side the greenhouses and radiated into space from theirroofs. To keep the heat transfer passive and thus reli-able, the radiator must be cooler than the greenhouse.At radiator temperature of + / m at emissivity of 0.9. Thusto dissipate 3.2 MW of power one needs 10560 m ofgreenhouse roofs. The greenhouses are stacked in asmany layers as is needed to yield the wanted amount ofradiated cooling power per roof area. The green areasin Fig. 1b show the greenhoused areas. Pressure con-tainment of the greenhouses also contributes to struc-tural mass. The mass is proportional to greenhouse totalvolume. To calculate the volume, we assume 50 W / m of volumetric power dissipation and 1 bar greenhousepressure.Agriculture is performed robotically because thegreenhouses are outside of the thick radiation shields.We make an assumption that agriculture works despiteGCR. To what extent this is true depends on many fac-tors, one of which is plant lifetime. Plants that growrapidly from seeds are less likely to have problems dueto radiation-induced mutations than long-lived trees, forexample. More research is needed on this point.We place the greenhouses so that they can be servedby the ringroads. The ringroads are used not only bypeople, but also by wheeled robots that move crops tothe living spheres and human wastes back to the green-houses. The indicated area of the cylindrical solar panelin Fig. 1b is larger (by factor of 2) than what is neededto produce 3.2 MW, because the settlement needs poweralso for other purposes than food production. The spinaxis is perpendicular to the ecliptic plan so that the so-lar panels are all the time optimally illuminated. Thepower system is a rather small fraction of the total mass,which is dominated by radiation shielding mass (97 %)and structural mass (2.6 %).Figure 2 shows a cross-sectional view of the spheres.The inhabitants live in the centrifugally produced ar-3 NSS Space Settlement Journal pin axis m Figure 2: Cross-sectional view. tificial gravity, with their heads towards the axis of ro-tation. Each sphere has inner diameter of 33 m. In thissection we give the baseline values of the parametersand motivate them later. With e ff ective room height of3 m it contains ten floor levels and provides 6000 m offloorspace area for its 100 inhabitants so that each per-son has 60 m of living area. For example, it can be 25m of private area per person, 32 m of public and work-ing areas including corridors, and 3 m (5 %) taken bywalls. Tables 2 and 3 list the main parameters of the 200person settlement. Table 2: Parameters of 200 person settlement.
Population 200Floorspace per person 60 m Sphere inner diameter 33 mRotation radius 230 mSteel tension 800 MPa (30 % of limit)Radshield mass per area 9.04145 t / m Radshield density 2.6 g / cm Radshield thickness 3.5 mFor relatively small settlements such as 200 inhabi-tants, the density of the regolith a ff ects the mass: thedenser the shield, the less of it is needed because thesphere’s radius is not that much larger than the shieldthickness. Bulk densities of asteroids vary in ratherwide range. Asteroid Eros is stony and with bulk den-sity of 2.67 g / cm [10]. Smaller stony asteroids are lesscompressed by gravity and they can have lower densi-ties; for example Itokawa has 1.95 g / cm . The miner-als themselves would allow even higher densities. Themost abundant minerals are SiO whose density is 2.65g / cm and MgO which is 3.6 g / cm . Iron-rich mineralsare often even denser. We use the value 2.6 g / cm whichis a bit less than Eros’ bulk density. Reaching this den- sity requires the fragmented rock to be compressed byvibrating, pressing or melting. Table 3: Power and mass budgets of 200 person settlement.
Total Fraction Per personPower 3.2 MW 16 kWRadshield 86423 t 96.9 % 432 tStructural 2270 t 2.55 % 11.35 tOther 493 t 0.55 % 2.47 tTotal 89186 t 100 % 446.0 tThe mass e ffi ciency of radiation shielding increases ifthe population is increased, because the spheres becomelarger. When making the spheres larger, however, thedi ff erence in artificial gravity between the top and thebottom increases, unless one also increases the rotationradius. When scaling up we require that the maximumgravity is not more than 10 % larger than the average,and the minimum is not more than 10 % smaller.Increasing the rotation radius increases the structuralmass fraction, because the sphere-supporting trusses be-come longer. We propose that the structural mass is pi-ano wire steel. This material is 99 % of iron, and ironis abundant on asteroids. In the first phase the struc-tural material is brought from Earth, but in later stagesit can be sourced from asteroids. We also set a require-ment that the structural fraction does not increase be-yond 17 %, because the majority of stony asteroids havemore iron than this limit. Nothing prevents an evenlarger iron fraction, but the only drawback is that thenone may have to start abandoning some of the asteroidmaterial as waste. When the rotation radius is increasedto 1.6 km, the 17 % limit is reached. If one wants tofurther increase the population without increasing themax / min gravity di ff erence beyond ±
10 %, one can addmore spheres. Configurations of up to ∼
30 spheres are4
NSS Space Settlement Journal till more mass e ffi cient than an uninterrupted torus.Figure 3 shows the rotation radius, mass per person,structural mass (steel) per person and the number ofspheres as a function of population. For each popu-lation, the mass-optimal configuration was found auto-matically. The rotation radius is the constant 230 m upto 600 people, after which it grows in order to avoidmore than ±
10 % di ff erence in gravity between spheretop and bottom. The mass per person is inversely pro-portional to sphere radius, so that it is proportional topower − / PopulationMass per person (tonne)Rotation radius (m)Number of spheresSteel per person (tonne)
Figure 3: Scaling as function of population.
With the employed parameters, two is the optimalnumber of spheres up to population of 2 · .Table 4 summarizes the employed requirements. Table 4: Employed limitations.
Parameter Value MotivationRotation rate ≤ ff erence in gravity ≤ ±
10 % Life convenienceStructural fraction ≤
17 % No wastedasteroid material
4. Material sourcing
Most of the material is asteroid rock or regolith usedfor radiation shielding. This fraction is 97 % in the base-line case of 200 people. Thus one can already reach as-teroid mass ratio of 30 [ = / (100 −
5. Propulsive transfer of materials
Most of the mass (97 % for a 200-person dwelling)is radiation shielding, for which there are no structuralor other requirements so that it can be any unprocessedor processed asteroid rock. Thus the main challenge ispropulsion: how to transfer material from asteroids orthe Moon to the L5 orbit. Here we briefly analyze someof the potential methods.
Lunar material could be lifted by an electromagneticmass driver as envisioned by O’Neill [11]. The massdriver would be a large investment. The electrical en-ergy of the shot must be stored in large capacitor banks,which is a major cost item. The fixed shooting directiontends to reduce the flexibility regarding target orbit. Thecentralized nature of the facility is a potential reliabilityconcern.Lunar material could also be lifted by a sling [12, 13],which would be much lighter infrastructure than theelectromagnetic mass driver and it avoids the use ofcapacitor banks. However, because the Moon rotateswhile the plane of the rotating sling stays inertiallyfixed, the sling crashes to the surface after some time,unless prevented by propulsion or other means. Thetime to crashing depends on the parameters, but is typi-cally inconveniently short.5
NSS Space Settlement Journal o lift lunar material, it is often proposed to makeLH / LOX propellant from the water ice that exists inthe polar lunar regions. However, the estimated H O re-source is only a few times 10 kg [14]. This amount isnot su ffi cient for long-term use. For example, a settle-ment with 10 people corresponds to mass of 3 . · kgper person (Fig. 3), i.e. total mass of 3 . · kg, whichis already ∼
10 % of the total lunar water resource . electricpropulsion As shown recently [15], the so-called FFC Cam-bridge electrolytic process can be used to separateearthly, lunar or asteroid rock into oxygen gas and asolid residue comprising metals and silicon. The oxy-gen can be used as Electric Propulsion propellant [16].The O Electric Propulsion technology is currently un-der development in the context of Air-Breathing Elec-tric Propulsion [16], which is enabling technology forvery low orbiting satellites that are naturally immune toorbital debris and do not generate new debris.The FFC Cambridge process requires calcium chlo-ride electrolyte. The electrolyte can be recycled, but theinitial amount must be brought from Earth. Chlorine isa rather rare element on asteroids. In the proof of con-cept experiment of Lomax et al. [15], 1.6 kg of CaCl was used to process 30 grams of lunar regolith simulant.According to the newest results [17], ∼
75 % of the to-tal oxygen was extracted after 16 hours in the reactor.Thus, during one year, 545 batches can be processed,altogether processing 16.4 kg of regolith and liberating4.3 kg of O , if the total oxygen content of the rock is35 %. Thus for one year, the mass ratio (O : CaCl ) is(4.3 : 1.6) = (2.7 : 1). Because the process demonstra-tion was intended only as a proof of concept and wasnot optimized, it is likely that the amount of electrolyteand / or the throughput time can be improved, maybe sig-nificantly. To estimate the delta-v from a given asteroid to L5,we compute the optimal Hohmann transfer delta-v fromthe asteroid to circular zero inclination 1 au heliocen-tric orbit, consisting of two or three impulsive burns,whichever strategy gives the smallest delta-v. The burnsset the aphelion, the perihelion and the inclination. Inreality, since we are considering low-thrust ElectricPropulsion, the burns are not impulsive and therefore The chemical propellant mass is of the same order of magnitudeas the payload mass in the lunar case. they are not optimal. On the other hand, in realityone could make use of lunar flyby maneuver to killup to 1 . / s of of the incoming hyperbolic excessspeed. Because the e ff ects work in opposite directionsregarding the needed delta-v, we think that the impul-sive Hohmann transfer delta-v gives a useful approxi-mate measure of the low-thrust delta-v.Figure 4 shows the cumulative mass in known aster-oids sorted by the delta-v computed as just explained.The masses in Fig. 4 are based on the tabulated absolutemagnitudes in JPL Small Body Database by assumingalbedo of 0.15, density of 2 g / cm , and spherical shape.The cumulative mass jumps at certain large and well ac-cessible asteroids, some of which are marked in Fig. 4.To build the first settlement, we need 78000 tonnes ofasteroid rock, which is only little larger than the es-timated mass of asteroid 2000 SG 344, which in onesource has been estimated as 7 . × kg [18]. To beconservative, however, we shall assume the use of as-teroid Apophis whose mass is 3 orders of magnitudelarger. Apophis is also one of the potentially hazardousasteroids, so reducing its mass is not harmful from theplanetary defense perspective.The delta-v from Apophis is 3.37 km / s. For the trans-fer spacecraft using O electric propulsion, we assumethe parameters listed in Table 5. Table 5: Parameters of the electric propulsion transfer spacecraft.
Delta-v 3.37 km / sSpecific impulse 2500 sE ffi ciency 0.4Traveltime 1.5 yearsAcceleration when thrusting 0.12 mm / s Fraction of thrusting arcs 60 %Power / mass of power system and thruster 100 W / kgTank mass versus tank content (LOX) 0.01 → Propellant mass versus rock mass 0.15 → Dry mass versus rock mass 0.043 → Mass ratio 23The assumed specific impulse of 2500 s is on the highend of Hall thrusters and low end of gridded ion engines[19]. The assumed e ffi ciency of 40 % is somewhat lowerthan the e ffi ciency of state of the art xenon Hall thrusterswhich are typically above 50 %. We motivate the as-sumption by the fact that O is less optimal propellantthan xenon.The power per mass ratio of 100 W / kg is typical tocontemporary solar panel power systems. The thrusteris typically quite lightweight in comparison. We assumea passively cooled LOX tank. The tank walls can be thin6 NSS Space Settlement Journal kg 0 1 2 3 4 5 6 7 8Delta-v from marginal escape orbit (km/s) In orbit nowBFR fleet LEO/yearLunar H O resource
Figure 4: Cumulative mass of known asteroids as a function of delta-v. since the pressure is low and the tank does not have towithstand launch vibrations or impulsive accelerations.We obtain the result that 13 % of the initial rock must beturned into oxygen, so that the ratio (propellant : rock)becomes (0.15 : 1). The 13 % corresponds to only aboutone third of the total oxygen content of typical asteroidrock.Thus, the ratio of payload to dry mass of the transferspacecraft is 23. If the spacecraft is used for multipletrips (1.5 year transfer followed by shorter triptime forgoing back to the asteroid), the e ff ective mass ratio isincreased. O electric propulsion
A drawback of the FFC Cambridge process is thatthe CaCl electrolyte must be imported from Earth. In-stead of extracting O from rock, one can use a (C-type)water-rich asteroid, extract water by heating, and usethe H O for Electric Propulsion. For Electric Propul-sion thrusters, O and H O are rather similar. Both arelight molecules and when in hot and ionized state, theyare chemically active. The amount of water in the material is likely to beless than what is needed for propulsion, unless the spe-cific impulse is chosen to be particularly high or theasteroid is particularly wet. Thus one must probablybe prepared for mining and drying more material thanone transports. To handle the waste material in the mostsustainable way, one can create an artificial asteroid ofthe abandoned material and set it to orbit the parent as-teroid. Then the dried-up material is not wasted per-manently, but it remains most easily accessible to thosefuture miners who prefer water-independent transporta-tion techniques such as FFC Cambridge.Water is easier to store than O because it is storableas a room temperature liquid. The Swedish-Finnish steelmaking company SSABintends to replace traditional carbon reducing agent insteelmaking by hydrogen, thus enabling CO -free pro-duction of steel. Hydrogen gas reacts with iron oxidesto form reduced iron and water. The water is electrol-ysed to extract the oxygen. The hydrogen is injectedback into the reactor.Ordinarily, hydrogen reduction is applicable only toiron oxides. Turning the hydrogen to atomic or ionizedform might also allow reduction of other oxides [20].This is relevant for asteroids because most of the oxygenis bound in silicon and magnesium oxides.The benefit relative to FFC Cambridge is that there isno need to import process chemicals from Earth. Hy-drogen is needed, but it can be circulated, and the initialamount can be obtained from the water that exists inmost asteroids to some extent. The solar wind electric sail (E-sail, [21, 22]) is a pro-pellantless propulsion method, based on the momentumflux of the solar wind. The E-sail consists of long andthin metallic tethers that are kept in high positive poten-tial by an onboard high voltage source and electron gunthat pumps out negative charge from the system. Themaximum thrust of an E-sail depends on materials andother parameters, but is roughly of the order of ∼ / s as in Table 5,the single transfer spacecraft can thus move 10 tonnesof material. To build a 200 person settlement weighing78000 tonnes (Table 2), one needs ∼ trips. Thus oneneeds a large fleet of E-sail spacecraft.The E-sails are tens of kilometers in diameter. Thus,tra ffi c congestion at the asteroid and at the settlementconstruction site become issues. The problem can be7 NSS Space Settlement Journal voided by moving the materials first by traditionalspacecraft which rendezvous with E-sails once thereis enough free space around. This increases the com-plexity to some extent, but the scheme remains e ffi cientsince the majority of the delta-v comes from the propel-lantless E-sail.A fully autonomous optical navigation system is pre-ferred. Otherwise operations and radio communicationcosts can become high, because the fleet is large. Anautonomous optical navigation system was in principledemonstrated already in Deep Space 1 in 1998.The E-sail tethers are made of multiple wires to betolerant of the natural micrometeoroid flux. However,centimeter-sized particles can break all the subwires ofa tether at once. Debris possibly generated by the aster-oid mining is thus potentially dangerous for the E-sails.Making rendezvous with the E-sails far enough from theasteroid also mitigates this problem. Table 6 summarizes the benefits of the four propul-sion options for moving asteroid materials.
Table 6: Comparison of propulsion options for moving asteroid mate-rials (Section 5). O H O H E-sailEP EP reduct.Any asteroid + ( + ) + No waste + +
No cryogenic tank + +
No Earth-imports + + +
Easily dockable + + +
No large fleet + + +
The Dutch-Luxembourgian company Maana Electric( ) is developing aself-contained automatic factory, built in a standard-sized shipping container that takes in desert sand roboti-cally and produces finished solar panel arrays, installingthem in the surrounding desert. The company targetsnot only Earth, but also the solar system. If this tech-nology proves to be practical, one could use it to pro-duce solar panels for the transfer vehicle from the minedasteroid regolith, thus reducing the mass that must bebrought from Earth. Structural parts of the transfer ve-hicle may be possible to 3-D print from the metal-richresidue of the FFC Cambridge process.Making parts of the transfer vehicle from asteroidmaterials is simpler than making parts of the settlement, because the transfer vehicles are unmanned and redun-dant. Thus a failure of one of them is not catastrophic.However, if strict quality checking standards are im-posed, then it becomes feasible to also make structuralparts of the settlement from asteroid materials. Ourbaseline structural material is piano wire steel. Pianowire steel is 99 % iron, which is abundant on asteroids.Other structural materials such as magnesium alloys orbasalt or silica fibers can also be considered.
6. Costing of 200 person settlement
Table 7 gives the masses and launch costs of the 200-person settlement. We assume FFC Cambridge (subsec-tion 5.2) and that the asteroid surface miners and the O extraction factory together can weigh up to 3000 tonnes.The contribution of surface miners is likely negligiblein comparison with the factory. In subsection 5.2 wefound that at the current unoptimized prototype level ofthe FFC Cambridge process, one unit of Earth-importedCaCl electrolyte produces 2.7 units of O per year.From Table 5, transportation of one mass unit of aster-oid rock needs 0.15 mass units of O propellant. Thus,transportation of one mass unit of asteroid rock needs0 . / (5 × . = , ifthe production period is taken to be 5 years. Thus, trans-portation of 96576 tonnes of rock needs 1062 tonnes ofCaCl , which is 35 % of the 3000 tonnes allowed for theO factory. Recall that this is based on the presently ex-isting unoptimized FFC Cambridge process of Lomaxet al. [17]. Table 7: Mass and launch cost budget for 200-person settlement, as-suming O propellant extraction but no other space manufacturing. Total Per pers. SourceAsteroid rock 86423 t 432 t Table 3Transfer s / c dry 3716 t 19 t Table 5Miner & O ex. 3000 t 15 t See textSteel 2270 t 11.4 t Table 3Other 493 t 2.47 t Table 3Earth to L5 9479 t 47.4 t SumEarth to LEO 28437 t 142 t 3 × L5Launch / F9 $85B $426M $3k / kgLaunch / Starship goal $853M $4.26M $30 / kgIn Table 7 we assumed – conservatively – that chem-ical propulsion is used to push the payloads from LEOto L5 so that the mass originally launched to LEO is 3times larger than the mass that ends up in L5. Falcon 9costs $3000 / kg to LEO in the default partially reusablemode. The fully reusable Starship rocket might be as8 NSS Space Settlement Journal uch as 100 times more cost-e ff ective ($30 / kg). Adopt-ing that, the launch cost per settler becomes $4.2M.Predicting the eventual per-kilogram cost of Starshipor its competitor is challenging at the moment. Onerecent statement of SpaceX speculated with only $2Mcost per launch, i.e. $13 / kg [23]. To cover the uncer-tainty, in the Discussion part we shall also consider anintermediate price case of $300 / kg.If the launch is e.g. 20 % of the total cost (the otherbeing designing and building the settlement and the as-teroid mining chain for obtaining the radshield rock),each settler needs an initial capital of 5 × . = $21M. Settlers can earn their investment back and more,since once living at L5, they can teleoperate the nearbyrobotic settlement production factory complex in zero-delay mode. Thus the first group of settlers earns moneyby producing new settlements. They do it more e ffi -ciently than from ground because they avoid the 2.6 sec-ond free-space communication delay in teleoperation.As was remarked above in subsection 5.7, solar pan-els and structural parts (including tanks) of the Elec-tric Propulsion transfer spacecraft could be made fromasteroid resources. Structural parts of the O factorycould also be be made of asteroid-derived steel. Onceproper quality control processes are in place, structuralparts of the settlement can also be made of asteroid-derived piano wire steel. In Table 8 we show an ex-ample where space manufacturing has cut the net Earth-imported mass to 20 % of the original. Table 8: Same as Table 7 but with space manufacturing.
Total Per pers. SourceMining & transfer 6716 t 33.6 t Table 7Steel 2270 t 11.4 t Table 7M&T e ff ective 1343 t 6.7 t 0 . × Steel e ff ective 454 t 2.3 t 0 . × Other 493 t 2.5 t Table 7Earth to L5 2290 t 11.5 t SumEarth to LEO 6870 t 34.4 t 3 × L5Launch / F9 $20.6B $103M $3k / kgLaunch / Starship goal $206M $1M $30 / kgThat is, under the stated assumptions space manufac-turing reduces launch costs by a factor of 4.2.
7. Discussion and summary
Radiation shielding dominates the mass of beyond-LEO settlements. The assumed 9 t / m shield providesmax 20 mSv / year equivalent dose environment. If the shield is made thinner, the e ff ective dose would increaserapidly.For large settlements where the shield thickness isnegligible in comparison with the sphere radius, themass density of the radshield does not matter. But for200-person settlements it does play a role. We assumeddensity of 2.6 g / cm which is the same as the bulk den-sity of larger asteroids such as Eros. Reaching this den-sity probably requires some technical e ff ort, for exam-ple making bricks out of it by pressing, sintering ormelting [24].It is advantageous to separate the water and to put itin its own layer, inward of the regolith. In this way, thehydrogen of the water moderates the spallated neutronsso that they can be captured by a thin layer of boron orother neutron absorber. A 2 % water content is su ffi cientfor moderating the neutrons. Water is also intrinsicallya better shield material than rock, so it is better if moreis available. To be conservative, in the calculations weassumed only 2 %.For moving the materials from asteroids there are sev-eral propulsion options. Extracting O from rock byFFC Cambridge and using it for Electric Propulsion is apossible method. A drawback is the necessity to importthe CaCl electrolyte from Earth. Finding a water-richasteroid and using H O as Electric Propulsion propel-lant is one of the other alternatives. The E-sail is alsoone option. It needs no propellant extraction, but re-quires a large fleet size.Tables 7 and 8 show the launch cost per person with-out and with space manufacturing, and with present(Falcon 9) and future (Starship) launch vehicles. Intro-duction of space manufacturing (under certain assump-tions) reduces launch costs by a factor of 4.2, and re-placing Falcon 9 by the Starship cost goal of $30 / kg re-duces them by a factor of 100 (Table 9): Table 9: Launch costs per settler with various launch prices and withor without space manufacturing.
LEO cost Space Permanufact. person$3000 / kg (Falcon 9) No $426M$3000 / kg (Falcon 9) Yes $103M$300 / kg No $42.6M$300 / kg Yes $10.3M$30 / kg (Starship goal) No $4.26M$30 / kg (Starship goal) Yes $1MIf Starship reaches $30 / kg, the first 200-person settle-ment could probably be built without space manufactur-ing. Each settler might need initial capital of ∼ $21M,9 NSS Space Settlement Journal / kg, which is 10 times smallerthan Falcon 9, but 10 times higher than the Starshipgoal. Because Starship launches 150 tonnes per launch,$300 / kg corresponds to each Starship launch costing$45M, which is about the same as the present Falcon9 launch cost in its default partially reusable configu-ration ($50M). Given that Starship is fully reusable, itslaunch should cost less than that of Falcon 9, becausethe fuel cost is only ∼ $2M per launch. In the $300 / kgcase, the launch cost per settler is $10.3M if space man-ufacturing is used. Such figure sounds realistic for 200settlers. If space manufacturing is not used, then thelaunch cost is $42.6M per settler. This initial cost levelis also probably feasible for 200 settlers, provided thatthere is a credible roadmap for lowering costs in the fu-ture by implementation of space manufacturing and / orby lowering launch costs.We summarize our main findings:1. The radiation shield against GCRs dominates themass (97 % in the case of 200 person settlement).It can be asteroid rock.2. A sphere is the optimal shape for radiation shield-ing. Hence the two-sphere dumbbell configurationis best.3. As structural material we recommend piano wiresteel.4. Several propulsion options to transport rock fromasteroid to L5 are viable.5. To cut costs by space manufacturing, one can makesolar panels and structural parts of the transfer ve-hicles and the settlement from asteroid materials.6. The economic case looks promising if LEO launchcost is $300 / kg or below.7. The settlers make money by constructing more set-tlements by teleoperating a nearby robotic factorywith negligible communication delay.
8. Acknowledgement
The results presented have been achieved under theframework of the Finnish Centre of Excellence in Re-search of Sustainable Space (Academy of Finland grant number 312356). I thank Thorsten Denk for making theauthor aware of the problem that the lunar sling has dueto the Moon’s rotation.
References [1] O’Neill, G.K., The colonization of space, Physics Today27, 9, 32–40 (1974). https://space.nss.org/the-colonization-of-space-gerard-k-o-neill-physics-today-1974/ [2] O’Neill, G.K., The high frontier, New York (1977).[3] Globus, A., S. Covey and D. Faber, Space settle-ment: an easier way, NSS Space Settlement J. (2017). https://space.nss.org/media/NSS-JOURNAL-Space-Settlement-An-Easier-Way.pdf [4] “Assure Space won’t cover collision risk in low Earth orbit”,Space News, March 11 (2020). https://spacenews.com/assure-space-leaves-leo/ [5] Hou, X.Y., X. Xin, D.J. Scheeres and J. Wang, Stable motionsaround triangular libration points in the real Earth–Moon sys-tem, Monthly Notices of the Royal Astronomical Society, 454,4, 4172–4181 (2015). https://academic.oup.com/mnras/article/454/4/4172/997896 [6] Globus, A., and J. Strout, Orbital space settlement ra-diation shielding, NSS Space Settlement J. (2017). https://space.nss.org/media/NSS-JOURNAL-Space-Settlement-Radiation-Shielding.pdf [7] Kramer, R., H.J. Khoury, J.W. Vieira, E.C.M. Loureiro,V.J.M. Lima, F.R.A. Lima and G. Ho ff , All about FAX: A Fe-male Adult voXel phantom for Monte Carlo calculation in radi-ation protection dosimetry, Phys. Med. Biol. 49, 23, 5203–5216(2004).[8] Hassler, D.M., et al., Mars’ surface radiation environment mea-sured with the Mars Science Laboratory’s Curiosity rover, Sci-ence, 343, 6169, 1244797 (2014).[9] Globus, A. and T. Hall, Space settlement populationrotation tolerance, NSS Space Settlement J. (2017). https://space.nss.org/media/NSS-JOURNAL-Space-Settlement-Population-Rotation-Tolerance.pdf [10] Yeomans, D.K., et al., Radio science results during the NEAR-Shoemaker spacecraft rendezvous with Eros, Science, 289,5487, 2085–2088 (2000).[11] O’Neill, G.K. and H.H. Kolm, High-acceleration mass drivers,Acta Astronaut., 7, 11, 1229–1238 (1980).[12] Baker, D. and R. Zubrin, Lunar and Mars mission ar-chitecture utilizing tether-launched LLOX, AIAA 90–2109,AIAA / SAE / ASME / ASEE 26th Joing Propulsion Conference,July 16–18, 1990, Orlando, Florida.[13] Landis, G.A., Analysis of a lunar sling launcher, J. British Inter-planetary Soc., 58, 9 /
10, 294–297 (2005).[14] Eke, V.R., L.F.A. Teodoro and R.C. Elphic, The spatial distribu-tion of polar hydrogen deposits on the Moon, Icarus, 200, 12–18(2009).[15] Lomax, B.A., M. Conti, N. Khan, N.S. Bennett, A.Y. Ganin andM.D. Symes, Proving the viability of an electrochemical pro-cess for the simultaneous extraction of oxygen and production ofmetal alloys from lunar regolith, Planet. Space Sci., 180, 104748(2020).[16] Andreussi, T., E. Ferrato, V. Giannetti, A. Piragino, G. Cifali,M. Andrenucci and C.A. Paissoni, Development status and wayforward of SITAEL’s air-breathing electric propulsion, AIAAPropulsion and Energy 2019 Forum, 19–22 August 2019, Indi-anapolis, USA, AIAA 2019–3995, doi:10.2514 / NSS Space Settlement Journal
17] Lomax, B.A., M. Conti, N. Khan, A. Ganin and M. Symes, TheMetalysis-FFC process for the e ffi cient extraction of oxygen onthe lunar surface, Presentation given in ESA Space ResourcesWorkshop, Luxembourg, October 10–11, 2019.[18] https://web.archive.org/web/20041024225737/http://neo.jpl.nasa.gov/risk/2000sg344.html [19] Goebel, D.M. and I. Katz, Fundamentals of elec-tric propulsion: ion and Hall thrusters, Wiley (2008). https://descanso.jpl.nasa.gov/SciTechBook/st_series1_chapter.html [20] Sabat, K.C., P. Rajput, R.K. Paramguru, B. Bhoi andB.K. Mishra, Reduction of oxide minerals by hydrogen plasma:an overview, Plasma Chem. Plasma Process., 34, 1–23 (2014).[21] Janhunen, P., Electric sail for spacecraft propulsion, J. Prop.Power, 20, 763–764 (2004).[22] Janhunen, P., P.K. Toivanen, J. Polkko, S. Merikallio, P. Salmi-nen, E. Hæggstr¨om, H. Sepp¨anen, R. Kurppa, J. Ukkonen,S. Kiprich, G. Thornell, H. Kratz, L. Richter, O. Kr¨omer,R. Rosta, M. Noorma, J. Envall, S. L¨att, G. Mengali,A.A. Quarta, H. Koivisto, O. Tarvainen, T. Kalvas, J. Kaup-pinen, A. Nuottaj¨arvi and A. Obraztsov, Electric solar wind sail:Toward test missions, Rev. Sci. Instrum., 81, 111301 (2010).[23] https://techcrunch.com/2019/11/06/elon-musk-says-spacexs-starship-could-fly-for-as-little-as-2-million-per-launch/ [24] Soilleux, R.J., Orbital civil engineering: waste silicates re-formed into radiation-shielded pressure hulls, J. British Inter-planetary Soc., 72, 244–252 (2019).11