Conceptual Designs for Volatile Mining Operations in Lunar Cold Trap Environments

44th International Conference on Environmental Systems ICES-2014-195 13-17 July 2014, Tucson, Arizona

Kyle Kotowick ,∗ David Barmore ,∗ Lynn Geiger ,∗ Thomas Coles ,∗ and Jeﬀrey Hoﬀman † Massachusetts Institute of Technology, Cambridge, MA, 02139, USA

Long-term lunar habitation is an extraordinarily expensive endeavor, but can be made substantially more feasible through in-situ resource utilization of volatile compounds. One of the most likely locations of such volatiles is in lunar cold traps (LCTs): crater interiors near the north and south poles that lie in permanent shadow due to the sun angle being constantly below the crater rim. With no sunlight reaching the surface, LCT temperatures lie in the 30-50 K range, allowing volatile compounds to permanently exist in solid states. The lack of cyclic sublimation leads to a possible accumulation of these volatiles, such as water and methane, over several billion years. Orbital instruments on LCROSS and LRO suggest that the upper layer of regolith within LCTs could contain up to 2-5.6% water by mass. The opportunity to extract signiﬁcant amounts of water from LCTs would have wide- ranging applications, including life support resources and rocket propellant for local use or export to Earth orbit. This paper describes a conceptual design for a robotic volatile mining operation within Shackleton Crater near the southern pole, one of the most promising locations for volatile accumulation, with a focus on designs for handling the harsh thermal and lightless environment of LCTs. Speciﬁcally, this paper provides a thermal and power requirements analysis, proposes a base/rover architecture, and explores the tradespaces for mining rovers, power production and transmission systems, navigation systems, resource export systems, hibernation (wintering) options, and initial set-up options. The proposed overall design is comprised of a feasible selection for each of these, with particular concern for integration, mobility, and scalability.

I. Introduction

One of the greatest barriers to long-term lunar habitation is that of resource requirements, which require a substantial and extravagant ongoing financial commitment. Continuous supplies of oxygen and water are required for human life support systems,1 but could also be used for greenhouses,2 electrical power production with fuel cells and a supply of hydrogen,3 or electrical power storage with a reversible fuel cell. The expense would therefore be reduced significantly if the transport of these resources could be performed at lower cost or even rendered unnecessary. It will now be shown that the key to reducing transport costs is reducing the requirement for launching propellant from Earth; this would not only be significant for a potential future lunar colony, but also for existing commercial and government interests in space. The fuel/oxidizer combination offering the highest specific impulse is hydrogen/oxygen, which is commonly used in rocket upper stages to transport communications satellites from a LEO parking orbit to GTO.4 Similarly, hydrogen/oxygen upper stages are often the best chemical propulsion choice for other in-space maneuvers, including trans-lunar injection and trans-Mars injection, as long as they occur soon after launch and hence do not require long-term storage of cryogenic propellants. However, although the high specific impulse means that the propellant mass is lower than for other fuel/oxidizer combinations, it is nevertheless generally the dominant contributor to the overall mass that must be launched from Earth for any of the missions just described. For example, the hydrogen/oxygen propellant mass for the Delta IV launch vehicle’s 4 m upper stage is 20,410 kg, which is much larger than the stage’s 2,850 kg dry mass or even the largest payload mass that it can place into GTO: 6,390 kg.5 The mass fractions on other hydrogen/oxygen stages are similar, as shown in Ref 6.

∗Corresponding Author, Graduate Student, Department of Aeronautics and Astronautics †Professor of the Practice, Department of Aeronautics and Astronautics

1 of 37

International Conference on Environmental Systems In light of this, the costs associated with both space exploration and geostationary satellite launches could be significantly reduced if most of the upper stage propellant did not need to be lifted out of the Earth’s gravity well, i.e. if the upper stage could be refuelled with propellant already in LEO (or even already in lunar orbit to return astronauts to the Earth after a lunar mission). It would then be possible to either carry a larger payload on the same launch vehicle or carry the same payload with a smaller first stage. The latter is particularly significant for the future of space exploration, as the NASA Space Launch System vehicle currently under development will be extremely expensive; this is not only because of its size and performance, but also because its missions will be relatively infrequent, resulting in a slow learning curve for efficient operation and a significant impact from fixed costs. These issues would be addressed by the use of a smaller existing commercial launch vehicle instead; such a vehicle would be more than sufficient to launch the individual components of an exploration architecture, but it would not be able to carry the large propellant mass required and hence would require refueling.7 Refueling is a considerable technical challenge, not least because cryogenic liquids must be stored for a considerable time before being offloaded. However, there has been significant work on propellant depots in recent years with concepts involving sun shields attached to lengthened ULA rocket upper stages.7, 8, 9, 10 Sun shields are not perfect, but cooling to lower-than-normal temperatures before launch is one approach to allowing for a long storage time before any significant propellant boil-off occurs.11 Even though there is existing interest in refueling to improve launch vehicle capability, the propellant would nevertheless still need to be transported from the Earth at considerable expense. The problem could be solved by transporting it from the Moon instead; it would still not be free, but the significantly lower gravity of the Moon would simplify the problem. With this as motivation, this report is concerned with the extraction of water from the Moon’s surface; it could then be electrolyzed to form hydrogen and oxygen propellant for transport to the depots described above. Water extraction is one aspect of in situ resource utilization (ISRU): the use of local (lunar) materials. In the context of a lunar colony, water extraction could serve to not only reduce transportation costs, but also to meet the colony’s water and oxygen requirements with local resources without transport from Earth.12 One recent study estimated that the use of ISRU would result in a reduction of the requirement for mass launched from Earth by up to 90%, when applied to both propellants and colony resources.13 The presence of water on the Moon was first suggested in 1961,14 but compelling empirical evidence first arrived in the form of neutron readings made with a detector on the Lunar Prospector spacecraft in 1998-1999.15 These neutron readings identified the presence of hydrogen, thereby indicating a high probability for the presence of water. Further evidence came with the analysis of the top 1-2 mm of soil by a spectrometer on the Chandrayaan-1 spacecraft16 and an impactor that it released. Final confirmation arrived when LCROSS impacted the permanently shadowed Cabeus crater17 and the accompanying LRO spacecraft provided detailed radar, altimeter, thermal, and neutron data to more precisely describe the water distribution.18 Note that earlier work had indicated the presence of blocks ice on the Moon by inference from radar measurements;19 however, this inference was later demonstrated to be incorrect, as the signal that supposedly corresponded to ice was in fact due to rough terrain, such as that found on crater walls.20, 18 Lunar Cold Traps (LCTs) are regions that are in permanent shadow, found in craters sufficiently close to the poles for the sun angle to always be lower than that of the crater rim; some cold traps have not seen the Sun for billions of years and have accumulated water ice deposits in that time.21 It is from these deposits that ice could be mined for use as propellant or in support of a colony. This paper presents a conceptual design for a lunar volatile mining project. It begins by describing the environment of the LCTs, followed by a general overview of an operation architecture and base layout. The proposed technical design is then described, starting with the rover, which uses JPL’s ATHLETE as a baseline. The ATHLETE rover’s abilities are assessed in light of the requirements of an LCT mining operation, with an emphasis on the thermal and power systems. This is followed by a discussion of the requirements for winter hibernation, along with the energy collection system to be employed. Finally, ideas for potential future work on this project are presented.

II. Lunar Cold Traps

II.A. Environment Both the north and south polar regions of the Moon contain craters with ﬂoors in permanent shadow from a highly oblique angle of insolation. Because the lunar axis of rotation is inclined by only 1.5◦ from a normal

2 of 37

International Conference on Environmental Systems to the ecliptic,22 the insolation does not vary much over the course of a year and so temperatures reach as low as 29 K in places, see Figure 1.21 The temperature in these craters is approximately 40 K on average, giving them the name Lunar Cold Traps (LCT). Cold trap areas cover roughly 5100 km2 in the south polar region and 2600 km2 in the north.

R Figure 1: Map of the average lunar surface temperature made using the Diviner instrument on the LRO spacecraft. The crater slightly oﬀset from the center of the axes is Shackleton crater. Image reproduced from Ref 21.

II.B. Volatiles Volatiles were originally deposited on the Moon by comet or asteroid impacts, or chemical reactions between oxides in the lunar regolith and protons deposited by the solar wind. Following deposition, volatiles can slowly migrate towards the poles through periodic heating and sublimation during the lunar day, followed by refreezing during the lunar night. Many of the sublimating particles reach escape velocity and are lost to the system, but slower particles follow a ballistic trajectory and return to the lunar surface. The process continues until the volatiles reach a location cold enough for them to remain frozen and stable, such as an LCT.23, 24, 25 The most valuable of the volatiles that will collect in these cold traps is water. There are many studies that attempt to characterize the amount of water that could be stored in LCTs, the most recent being the LCROSS mission which has calculated up to 5.6% water by mass in the upper 2-2.5 m of regolith inside Cabeus crater17 and a combined study of the various data collected from LRO, which indicated at least 2% water in the form of pellets or rock coatings, but not solid blocks of ice.18 Berezhnoy et al. believe that water ice could be stable in a 72,000 km2 area of the southern polar region,23 when including ice at any depth down to 1 m below the surface in areas outside of LCTs.21 After all, Figure 1 indicates that even some of the lunar surface outside the craters has an average temperature below 100 K. As explained in section IV.D.5 and,26 water ice sublimation is very slow, even at temperatures higher than 100 K, especially when covered by at least 2 cm of regolith. Figure 2 indicates the regions of stability for ice, including signiﬁcant areas outside LCT craters when considering ice below the surface.

II.C. Shackleton Crater The most studied LCT is Shackleton Crater and hence it will be the focus of this investigation. Shackleton crater is located in the Aitken basin, which is one of the largest and oldest impact craters on the Moon.22 The rim of Shackleton Crater is located exactly on the lunar south pole, making it quite unique - this can

3 of 37

International Conference on Environmental Systems Figure 2: Map of the depths below the surface of the lunar south pole at which there is long-term stability of water ice. This is inferred from a model that used the surface temperature data collected by the Diviner instrument on the LRO spacecraft. The white regions are LCT craters and the grey regions do not permit ice trapping within 1 m of the surface. Image reproduced from Ref 21, which also contains details of the model.

be seen in Figure 1. The crater is 21 km in diameter and 4.1 km deep.18 The walls of the crater have an average slope of 30◦ and a maximum slope of 35◦.18 The crater is 3.6 Gyr old, based on crater counts, and has been collecting volatiles in permanent shadow for 2 billion years, since the rotational axis of the Moon shifted to its present location.22 The annual average temperature range seen within Shackleton crater is 25-75 K, depending on the precise location. The maximum seen at any point within the crater during the year is 95 K.

III. Mining Operation Architecture

This project is based on the idea of centralized mining operation, from which mining rovers will get power, and to which all collected regolith material will be brought for refinement. An alternative approach would be decentralized mining, where each rover is equipped with refinement units and independent power supplies. This approach was deemed inefficient due to the large increase in rover weight required for the additional subsystems. This section encompasses the layout of the operation, the initial set-up process, and options for exporting the extracted volatiles.

III.A. Layout Mining rovers will receive power via a central processing plant, located on the floor of Shackleton Crater. Power will be transmitted to the plant using lasers from one or more towers located on the crater rim. The towers will be outfitted with solar panels to collect energy, a high-powered laser, and radio receivers to triangulate positions of rovers within the crater. The configuration of the towers is determined by the optimal placement to maximize yearly sunlight, as discussed in section V.B. A map of the proposed base layout is shown in Figure 3.

4 of 37

International Conference on Environmental Systems Figure 3: Schematic of proposed base layout shown on an elevation and slope map. Adapted from Ref 18.

III.A.1. Central Units The central processing unit will be comprised of two self-contained modules: a electrolysis module and a refinement module. The refinement unit is not necessary for the mining operations until enough raw regolith material has been collected and delivered to the refinery site. By refining the regolith inside the crater, the volume of material exported will drop 94-98%, as the estimate of water content ranges from 2-5.6%18.17 The electrolysis module will have a monochromatic-photovoltaic cell, which will be powered by laser energy transmission from solar towers on the crater rim. It will also have large storage tanks to keep the liquid H2 and O2 for refueling the mining rovers. Using a regenerative fuel cell on the electrolysis unit to supply electricity to the rovers through inductive energy transfer was investigated, but deemed unnecessary. The modularity of the central base will make it very portable, allowing for relocation if the ice distribution in the initial region is found to be unfavorable; this would decrease the rover traverse time, increasing the amount of time spent excavating.

III.B. Initial Set-Up The mining project will be installed in three phases. In the first phase, the power towers, fully assembled with the laser transmitter and solar panels, and a maintenance rover will land outside the crater. The solar-powered maintenance rover will utilize the long summer days to place the towers in their designated positions and assist with solar array deployment. Once all the towers are initialized, the rover will only be needed for maintenance of the towers and can be sent out for scouting or scientific purposes in the area. During the lunar winter, the rover will retreat to a nearby hill, which has been identified as a “peak of eternal light”,27 for power. This hill is also the location for the tower solar panels and transmitter, as discussed in section V.B. For the second stage of initialization, the payload will land directly inside the crater. This payload will consist of the ATHLETE rover, as described in section IV.A, the first half of the central base - the electrolysis unit, and enough fuel to sustain operation through the first winter hibernation. Once the rover has installed the electrolysis unit in the desired location, the rover will begin to gather regolith. It will bring the regolith to the central processing station, creating mounds to be processed in the final stage. The electrolysis unit will continually convert water produced by ATHLETE’s fuel cells back into H2 and O2. During the winter, the mining operation will halt and the rover will divert all energy to warming. In the spring, the final segment of the mining operation will be delivered. This will consist of a regolith refinement unit, a volatile exportation mechanism, and any additional ATHLETE rovers. This shipment could be landed either inside the crater or on the crater flanks, as the ATHLETE rover is capable of

5 of 37

International Conference on Environmental Systems maneuvering up to 30◦ slopes and the electrolysis unit will be set up to refuel the rover after the traverse. The refinement module will be attached to the electrolysis module and begin to refine the mounds of regolith collected during the previous year. By breaking the initialization into several steps, the weight of any one launch is decreased, thereby allowing flexibility in launch vehicle selection. An earlier launch also decreases the probability of canceling the project due to funding cuts. Stage two could run indefinitely before the refinement unit arrives, creating flexibility in the project schedule.

III.C. Volatile Evacuation Once the base has been established and volatiles start being collected, it will be necessary to transport these volatiles out of the crater. Three possible methods of accomplishing this are: using hoppers to make short jumps in and out of the crater, launching projectiles of mined material using hypervelocity accelerators, or building a cable lift system along the crater wall. Each of these options will be discussed in more detail below, the most favorable option at this time is cable winch, given its low energy requirements and high reliability. The shortest traverse up the crater wall, from the relatively flat crater floor to the crater rim, has been identified as 6 km in length and 3 km in elevation, and is depicted in Figure III.C.

Figure 4: Slope map of Shackleton Crater. The shortest traverse up the wall of Shackleton Crater is shown in black. Adapted from Ref 18.

III.C.1. Hoppers A lunar hopper could be used to export volatiles from the crater. The technological development of a lunar hopper has been slow, but the most complete lunar hopper project so far has been the Terrestrial Artiﬁcial Lunar And Reduced Gravity Simulator (TALARIS) developed by MIT for competition in the Google Lunar X-Prize.28 The TALARIS project used fans to counteract Earth’s gravity and simulate a lunar environment, along with cold gas jets as the propulsion proxy for a hydrazine fuel system.29 The project focused mainly on GNC algorithms and has yet to be tested outside a controlled environment. Because the objective for the hopper was to jump 500 m horizontally, it would be impractical to scale the capability of the TALARIS hopper design to export volatiles in this LCT setting: the TALARIS design only called for a vertical hop of 5-10 m and the energy required for a hop out of a crater is 600 times greater. Assuming the lunar hopper behaves like a rocket, the energy and fuel requirements can be calculated in the following manner:30 1 E = (m − m ) V 2 (1) rocket 2 0 f e m0 ∆V = Ve ln (2) mf

6 of 37

International Conference on Environmental Systems Where m0 is the initial mass, mf is the ﬁnal mass, ∆V is the change in velocity from the rocket burn, and V is the exhaust velocity calculated from V = I g, with g as the gravitational constant. ∆V is obtained e e sp √ 2 H+ H2+L2 using V = glunarL tan θ, where θ is the optimal launch angle calculated from tan θ = L . In the case of Shackleton Crater with the shortest traverse described above, H = 3 km and L = 6 km, yielding a θ of 58◦ and a ∆V of 125 m/s. The TALARIS hopper was designed to utilize hydrazine as a propellant, which has a vacuum Isp of 230 seconds.31 Using these values in Equation 2 yields a mass ratio of 1.057. Because the lunar hopper needs a soft landing, the fuel mass must be calculated in two parts; the fuel required to accelerate out of the crater and fuel for the braking and landing. A payload of 300 kg, comprising 100 kg for the hopper and 200 kg for mined material, requires an intermediate mass of 317 kg, which in turn requires an initial mass of 335 kg, which means 35 kg of hydrazine is needed for each load. The energy required to export 200 kg of goods out of the crater would be 89,400 kJ. Due to the complexity of a lunar hopper, the slow progress in hopper technology, and the very large amount of energy required, lunar hoppers are not a viable exportation option for this operation.

III.C.2. Hypervelocity Accelerators Rail-based hypervelocity accelerators are composed of two parallel conducting rails, with a conducting projectile touching both rails. A current is then sent down one rail, across the projectile, and back up the second rail. This creates a magnetic field, which accelerates the projectile very quickly, until it leaves the rails and enters a ballistic trajectory. These accelerators have been researched previously as a possible method to transport material from the lunar surface to orbit.32 Since the energy required to reach orbit is significantly higher than the energy required to reach the top of a crater, it should be relatively easy to scale down this technology for the problem at hand. The hypervelocity accelerator described in Ref 32 can accelerate a 2 kg projectile at approximately 2000 m/s2, has a linear mass density of 200 kg/m, and has a total efficiency of approximately 33%, with half of the remaining energy being transmitted as thermal energy to the projectile, and the other half as thermal energy to the accelerator. In order to launch a projectile over a horizontal distance L and a vertical distance H, the optimum launch angle θ, which minimizes both the energy required for launch and the velocity of the projectile when it lands at the top of the crater, is given by the expression: √ H + H2 + L2 tan θ = L This angle gives a launch velocity as follows:

2 v0 = gL tan θ where g is the local gravitational ﬁeld. The minimum track length to achieve this velocity, at acceleration a is:

2 v0 2a Finally, the change in temperature of the projectile is given by ∆Q ∆T = mCp If each mined water pellet is covered with only a thin layer of conducting material, it can be approximated as 2 kg pellets of pure water-ice, Cp = 2.108 kJ/kg·K. As previously mentioned, the shortest path from crater ﬂoor to rim occurs at H = 3 km and L = 6 km. Substituting these values into the above equations yields an optimal launch angle of 58◦, a launch velocity of 126 m/s, an impact velocity of 78 m/s, and an energy input requirement of 47 kJ per pellet. It also requires a track 4 m long with a mass of 800 kg and the resulting friction increases the average temperature of the projectile by 3.7 K during launch. The rails could be built onto the side of the crater, reducing the elevation relative to the ground to approximately 30◦. The energy could be sent from solar panels along the rim of the crater, and slowly stored

7 of 37

International Conference on Environmental Systems in a capacitor (or possibly a superconducting energy storage system, given the low crater temperature), which could then be quickly discharged whenever a pellet is ready for launch. Additionally, since the vacuum sublimation temperature of water-ice is over 100 K higher than the average crater temperature (see section IV.D.5), there would be little danger of losing much water heating the pellet during launch. The mass of the entire track would be slightly more than half a mining rover (described in the following section). The primary difficulty with hypervelocity accelerators is catching the projectile at the top of the crater. At a minimum velocity of 78 m/s, the impact would likely either bury the projectile deep within the regolith or destroy it on impact, which would make recovery of the volatiles difficult to impossible. Additionally, the long distance of the launch and the high speed of the projectiles means that precision is critically important. A tiny error in launch velocity could cause the projectile to miss its target, and if it hit part of the mining operation at the rim by mistake, the damage could be catastrophic. Although the lack of atmosphere on the moon would help mitigate aiming difficulties, it would still need to be addressed if hypervelocity accelerators were selected. Unless some method can be found for slowing the deceleration of the projectile when it lands, and aiming the projectile with pinpoint accuracy, hypervelocity accelerators will not be a viable alternative.

III.C.3. Cable Systems An alternative method for transporting material would be to use a cable system, such as a cable lift to transport material while suspended, or a winch system, which would drag the resources along the ground. A cable lift system is plausible, as cable lifts of similar size exist on Earth. The Tianmen Shan cable car in China is 7.5 km long and has sections with a slope of 37.8◦, which is greater than the slope along Shackleton’s walls.33 This cable car is however not quite comparable to the one that would be required at Shackleton, since the elevation change is only 1.3 km (compared to 3 km at Shackleton). Lift towers would need to be installed along the unstable crater wall for the elevated system. In terms of resource costs, using a winch is more viable than an elevated cable way, as a winch only requires a hoisting mechanism at the top of the crater and a single length of cable, half the length of a continuous loop system. A rough estimate of the power required to operate the winch to remove volatiles out of the crater is as follows. A steel cable of 1/4 inch diameter is capable of supporting 3 tons in Earth gravity34 and has a length density of 0.156 kg/m. For this application, approximately 6.7 km of cable is required, which would have a mass of 1045 kg. To tow a payload of mass mp up the crater wall, the change in potential energy can be calculated with the following equation:

h ∆U = g m h + m lunar p cable 2 If the payload were 300 kg (assuming 100 kg for the container and 200 kg of mined goods), the change in potential energy would be 3,900 kJ. Because this task is very simple, and using an ATHLETE rover is overly complexity, it would be more efficient use a simplistic cart design. Though the container would on treads or wheels, the worst case scenario for friction (dragging without wheels) of the 300 kg payload on a 30 degree slope with coefficient of friction of approximately 0.435 produces a force of friction of 90 N. Over 6.7 km, this results in an additional energy requirement of 600 kJ. Ideally the tow cable would remain taut and not have contact with the ground, more research into cable sag in lunar gravity is required to understand any additional energy lost to friction. The lowest allowed efficiency for an electric motor is 78.8%, according to NEMA (National Electrical Manufacturers Association) standards.36 This would produce an energy requirement of 5,700 kJ. However, the total efficiency of a drive system, including gears, may be as low as 50% at sub-optimal temperatures.37, 38 This efficiency would bring the total energy required to 9,000 kJ, which is much lower than the hopper and hypervelocity accelerators. The energy for this volatile extraction method could be supplied by solar panels directly powering the hoist, which is more efficient than beaming power into the crater to use the hypervelocity accelerator, as explained in section VI.B. The winch system is also safer, as the mined goods will not be traveling at high speeds at any time. The difficulty of this method lies in transporting 1045 kg of steel cable to the Moon, but there is a possibility for the cable to be created in situ using lunar iron deposits.

8 of 37

International Conference on Environmental Systems IV. Mining Rovers

Automated rovers will be used to harvest regolith from the crater floor and then transport it to the central refining station. If the mechanism chosen to transport the volatiles out of the crater is not located at the central station, the rovers will also transport the refined material to the location where it will be sent out of the crater. The rovers should be able to travel relatively long distances and carry large payloads to reduce the number of mining trips required. Ideally, these rovers will also be flexible enough to assist in setting up and relocating the base, as needed, rather than requiring specialized construction rovers for these tasks.

IV.A. ATHLETE One aim of this project is to use as much existing technology as possible in order to reduce development costs; this led to the selection of the All-Terrain Hex-Legged Extra-Terrestrial Explorer as the basis for the mining rover in this project. Developed by JPL speciﬁcally for the establishment and operation of a lunar base,39 ATHLETE already meets many of the design requirements.

IV.A.1. An Overview of ATHLETE’s Development and Specifications There are three planned iterations of the ATHLETE rover design, each roughly doubling in size from the previous one. The first version, completed in 2005, consisted of a single six-legged robot approximately 2 m tall, with a mass of 850 kg,40 and was estimated to be able to carry approximately 4500 kg of payload in lunar gravity.41 The second version, completed in 2009, consisted of two three-legged robots working together. Together, these Tri-ATHLETEs stand 4 m tall, have a combined mass of 1440 kg,39 and are estimated to be able to carry approximately 7,000 kg in lunar gravity.41 The third and final version of ATHLETE is still in the design phase; when it is finished, it should be 8 m tall with a mass of 2340 kg and a payload capacity of 14,500 kg in lunar gravity.42

Figure 5: Two Tri-ATHLETE vehicles prior to docking to the mobile habitat/pallet. Image reproduced from Ref 43.

Since the second version has been built and tested, while the hypothetical third version has yet to be designed, the rest of the analysis is based on the assumption that only the current Tri-ATHLETE version is available. If the proposed final version of ATHLETE is ever completed and demonstrated, it could only improve the capabilities of the operation, so this analysis can be seen as a lower bound on the effectiveness of mining LCTs for volatiles. The ATHLETE rover consists of six identical legs arranged around a central hexagonal body, which houses the electronics, power, and other payloads for the rover. These legs are designed to be as modular as possible: each leg is identical to the other five,43 and there are only two unique joint types for the seven joints on each leg.43 Thus, fewer spare parts are needed for repairs, and repairs will quickly become easier through experience with the limited number of components that could potentially fail. The legs and actuators can

9 of 37

International Conference on Environmental Systems also be used for mining, as seen in Figure 7; this reduces the mass requirement for mining-specific equipment on the rover, leaving more room to store regolith for processing. The ATHLETE rover is designed to travel at speeds of up to 5 km/hr on hard, flat terrain, but in field tests designed to roughly simulate lunar terrain, it usually travelled at less than 1.5 km/hr.44 If it could maintain this speed, it would take approximately 12 hours to travel from the central processing facility to the rim of the crater and back. However, this speed was on soft but relatively flat terrain; on the steep crater walls, it would be much slower, resulting in a much longer transit time, and consequently a need to carry more fuel to complete the journey (section IV.E). Therefore, it was decided that the refined volatiles will be transported out of the crater using one of the methods described in the previous section, rather than being carried out by the rovers.

IV.A.2. Beneﬁts of the ATHLETE Conﬁguration

Figure 6: Comparison of the eﬃciency of the ATHLETE rover architecture to the traditional rocker-bogie design. The green line indicates the mass of a loaded Tri-ATHLETE rover. Image reproduced from Ref 41.

An ATHLETE’s legs have the ability to roll on flat terrain, or walk/climb over rougher terrain. This added functionality actually reduces the mass of the entire rover, as the wheels do not need to be as robust as those on a traditional rover;41 the ground pressure can be increased to four times that of a rocker-bogie system, as a stuck wheel on ATHLETE is not usually a problem - it can simply walk out of difficult terrain. As can be seen in Figure 6, the rocker-bogie design is more efficient for small rovers, while the ATHLETE design is more efficient for large rovers. The Tri-ATHLETE design point is indicated in figure and is clearly a significant improvement over a standard design, even when just considering total payload capacity. This advantage is increased further by the use of ATHLETE’s limbs for mining without additional actuators, as discussed earlier in section IV.A.1 and shown later in Figure 7. Additionally, due to the generic and modular nature of ATHLETE’s design, these rovers can easily perform other tasks. Such tasks could include setting up the base, relocating the base to another part of the crater once the initial area is fully mined, or even moving the base to a new crater if the resources in Shackleton are ever completely exhausted. The latter would create significant difficulties in thermal management, as rovers and base components designed to operate permanently in a lunar cold trap may not survive at the warmer temperatures encountered during the journey; one way to be mitigate this would be to move during a lunar night.

IV.B. Electrical Power Requirements An important ﬁrst step in both the electrical and thermal analyses of the rover is to determine the power requirements while the rover is operating. Data available from NASA45 indicates that the avionics power requirement is approximately 500 W. During a typical traverse, the actuator power for all legs is approximately 4 kW, consisting of 1 kW for joints and 3 kW for wheels. Note that six of the seven joint actuators are active during traverse because

10 of 37

International Conference on Environmental Systems of the lack of suspension and resulting small joint movements as the rover moves over rough terrain; these movements must be compensated by the actuators, which also act to maintain the central body in a horizontal attitude.46 Due to the 60% eﬃciency of the fuel cell, as described later in the Power System section, the total thermal power that must be dissipated from the rover’s body is then approximately 3.2 kW.

Figure 7: The use of a scoop on one of ATHLETE’s legs for collecting regolith; note that no additional actuators are required beyond those needed anyway for mobility. Images reproduced from Ref 47.

Aside from traverse, the other main operating mode for the rovers is mining. This is shown in Figure 7 and is accomplished by lifting one of the legs, attaching a scoop, then collecting and lifting regolith to dump it into a container on top of the rover. The power requirement for mining depends upon the following: • The scoop volume, estimated from images in Ref 47, is approximately 0.05 m3. Assuming a regolith density of 1660 kg/m3, this corresponds to a regolith mass of 83 kg. Other relevant scoop properties are a height of approximately 30 cm and a mass of approximately 23 kg, again approximated from images in Ref 47 while using a wall thickness of 1.25 cm and assuming construction from aluminum 6061 with density 2700 kg/m3. • The mass of each ATHLETE limb is approximately 118 kg, estimated using Figure 6. This includes the wheel mass of 15 kg.43 • The height through which regolith must be lifted is approximately 2 m. This includes an 0.3 m indentation of the scoop into the regolith, 0.2 m for the rover’s main body, up to 1 m for the height of the regolith container, and an extra 0.5 m for the elevation of the rover body above the ground. Note that the body elevation number is conservative, as the rover’s other legs can lower it very close to the ground to improve mining eﬃciency. • Soil pressure at diﬀerent depths in simulated lunar regolith is described in Ref 48 and shown in Figure 8. Using a piecewise linear approximation to this plot, the total energy required to dig the scoop into the soil to a depth of 30 cm (the scoop’s height) is 8.5 J, which equates to negligible power usage when spread over a minute or even just a few seconds. • Judging from ATHLETE motion in NASA videos,45 the time for each scoop to be collected and deposited in the container would be approximately 2 minutes. Due to the relatively long time of 2 minutes for each scoop, the overall power consumption for mining – both lifting and scooping – is less than 10 W, even when assuming for simplicity that the entire limb mass must be lifted through the full height. This is negligible compared to the power requirement during traverse.

IV.C. Low-Temperature Materials Although any rover sent to the Moon must be capable of surviving extreme temperatures by Earth standards, the temperatures inside lunar cold traps require even closer attention to thermal considerations. This

11 of 37

International Conference on Environmental Systems Figure 8: The increase in soil pressure with depth is shown here for 4 test penetrations of simulated lunar regolith. Image reproduced from Ref 48. section provides an overview of different categories of materials for use in these cold temperatures; this is in preparation for section IV.D, which contains thermal calculations for an ATHLETE-style rover. The literature on prior experience in relevant temperature regimes is for applications that would normally involve cryogenic liquids - either the transport/storage of those liquids or uses such as particle physics experiments, infrared telescopes, and hydrogen-fueled rockets. Information on more practical cold temperatures, for example in the arctic, is of limited relevance, as many serious materials issues do not become apparent until much lower temperature than this (for example, see the polymer temperature ranges mentioned later); furthermore, many of the most significant challenges found in cold environments on Earth are related to freeze-thaw cycles of water rather than to the temperature itself. It is also worth noting at this point that the lunar cold trap environment is less stressful than many other space environments in one key aspect: the temperature is very stable, so thermal cycling (at least due to external influences) is not a concern once a system is in place inside a crater.

IV.C.1. Metals Metal crystal structures generally take one of three forms: Body-Centered Cubic (BCC), Face-Centered Cubic (FCC), or Hexagonal Close Packed (HCP). These are demonstrated in Figure 9. The arrangement of FCC crystals is such that atomic planes can slip more readily than in the others; FCC crystals are therefore the most ductile. Those with an HCP structure have the least number of slip planes and are typically less ductile than those with FCC structure, but can be more ductile than BCC because of the closer packing of atoms.49 As the temperature decreases for any crystal structure, the atoms move closer to each other and so slip requires more energy. This makes materials stronger, but also less ductile, as slip may no longer be possible at all in some directions. Most FCC metals nevertheless remain ductile because of the large number of possible ways of slipping. However, HCP metals and particularly BCC metals are generally brittle at these temperatures, hence they cannot be used for any supporting structures where tough materials are required. Despite this, appropriately chosen alloys and heat treatments for BCC or HCP metals can result in a product that is usable at these temperatures by mitigating the lack of slip problem in the original material, sometimes

12 of 37

International Conference on Environmental Systems (a) FCC (b) BCC (c) HCP Figure 9: The three diﬀerent crystal structures in which metals are typically found. FCC metals usually remain ductile down to cryogenic temperatures, but are not as strong as the others. BCC metals are always brittle at cryogenic temperatures and often even at moderately cold temperatures found naturally on Earth. HCP metals are more ductile than BCC, but are also often brittle at cryogenic temperatures, with the notable exception of low-interstitial titanium. by changing the crystal structure - for example, iron exists in many forms, including the common α (BCC) form and the γ (FCC) form. It is also possible to improve ductility at low temperatures for some otherwise relatively brittle metals by using a low-interstitial form, i.e. a form with a low number of contaminants or other disruptions to the lattice structure; this is because some atomic slip is then possible, but only if the lattice structure if largely free from distortions. In fact, the toughness of low-interstitial titanium is improved to the extent that a low-interstitial titanium alloy is used cryogenically in some parts of CERN’s LHC because of its exceptional strength at those temperatures.50 Although some cold metals are too brittle to be used for anything because they cannot withstand thermal stress during cooling (see, for example, some of the titanium alloys in Ref 51), other metals are still usable despite being brittle; of course this restricts the range of applications, but sometimes they have other desirable properties, such as with the use of the HCP metal beryllium in the mirrors of the 40 K James Webb Space Telescope.52 In general, FCC metals will be of interest for the lunar cold trap rover. Some common metals included in this category are: aluminum, copper, nickel, gold, silver, lead, and the gamma allotrope of iron, known as austenite. The properties of these metals are in many ways enhanced at cryogenic temperatures; the exact values depend upon the speciﬁc metal and alloy, but some representative values are given below:53

• Yield tensile strength increases by 15-50% and fracture tensile strength by 50-100% between 295 K and 20 K. This helps to compensate for the relatively low 295 K strength of these metals compared to non-FCC titanium alloys. • Young’s Modulus increases by 6-20% between 295 K and 20 K. • Thermal and electrical conductivity change dramatically, rising by more than an order of magnitude between 295 K and 20 K. • The thermal expansion coefficient drops by more than an order of magnitude between 295 K and 20 K, with most of the drop occurring below 100 K. This means that material deformation concerns become much less significant at temperatures below 100 K than above it. • The specific heat follows a similar pattern to the thermal expansion coefficient. This means that the temperature drops rapidly as heat is lost below 100 K; however, the benefit from this is that the power

13 of 37

International Conference on Environmental Systems required to raise the temperature from 20 K up to 100 K is far less than that required to raise the temperature by 80 K when starting from a higher temperature.

Note that some alloys do however yield significantly different properties and are sometimes not recom- mended for use at cold temperatures. Aluminum 7075-T6 is one such example51 and this is the material used in the existing version of Tri-ATHLETE.43 Initial work on modifying the rover considered survivabil- ity if the structure were to reach cryogenic temperatures. 7075-T6 is a particularly strong alloy at room temperature alloys and so looking at room temperature properties alone would not be sufficient to find a better alternative. For example, aluminum 2024-T4 has 27% lower yield strength than 7075-T6 and 12% lower ultimate tensile strength. However, the improvements to mechanical properties at low temperatures are beneficial here: the yield and tensile strengths of 2024-T4 at 78 K are superior to the room temperature values of 7075-T6. Below 78 K, 2024-T4 is superior to 7075-T6 at room temperature. 7075-T6 also improves in strength significantly as the temperature decreases, but this is not as helpful when it loses ductility. How- ever, 2024-T4 remains ductile even to low temperatures and is in fact sometimes even used to store liquid helium. In conclusion, an ATHLETE rover designed to use 7075-T6 aluminum at room temperature will exhibit similar or superior mechanical performance if constructed from 2024-T4 aluminum and operated at lunar cold trap temperatures. However, as demonstrated in the thermal analysis in section IV.D below, the rover itself will not actually be at such cold temperatures after all.

IV.C.2. Polymers Polymers are of interest for uses such as seals and insulation. All elastomers tend to become brittle at temperatures below ∼200 K, completely losing their elastic properties and easily fracturing. Most other polymers are similarly unsuitable for use at low temperatures. At higher temperatures, the elasticity of elastomers arises from the ability of polymer chains to easily slide past and unwind within the material. However, after passing through a glass transition temperature, these chain structures harden into an amorphous glass state and become brittle.53 There are some exceptions to this, as some polymers are highly (> 90%) crystalline and remain ductile below their glass transition temperatures; this is because their ductility comes from atomic slip planes within crystals rather than from sliding polymer chains. PTFE is the best example of such a polymer. Despite having a glass transition temperature of 388 K, it retains high toughness down to close to absolute zero. Two other suitable polymer categories at lunar cold trap temperatures are UHMW-PE (Ultra High Molecular Weight Polyethylene) and polyamides, which have successfully been used to create aerogels for insulation. Finally, polyurethane foam and Nomex are key materials used in the development of cryogenic insulation materials at NASA;54, 55 these are intended for use on liquid hydrogen tanks at even lower temperatures than those expected in lunar cold traps. The two main areas in which polymers are used on the ATHLETE rover are the wheels, as discussed later in section IV.D, and seals for the actuators. These polymers also have very low thermal conductivity53 and hence are suitable for use as insulators.

IV.C.3. Actuators and Lubricants Although polymers and metals capable of surviving lunar cold trap temperatures are readily available, unfortunately the same is not true of lubricants, which are the limiting factor for actuator operation. Liquids that are effective lubricants at room temperature on Earth typically become too viscous for use by 200 K, with some exceptions being usable down to 170 K - far above lunar cold trap temperatures.56, 37 As familiar lubricants are impractical, the first alternative to consider is cryogenic liquids that might be able to provide lubrication. When cryogenic liquids are pumped between storage facilities and transport containers, the pumps are often completely submerged in the liquid. Although the liquids are stored under high pressure, initial condensation typically occurs at 1 atm and so the 1 atm boiling point values are valid to establish approximate temperatures here.57, 58 The cryogenic liquids themselves, such as methane (boiling point 109.15 K at 1 atm) and argon (boiling point 87.35 K at 1 atm), must provide lubrication to the submerged pumps. However, their viscosity values are generally low, for example, 1/7 of that of water in the case of liquid nitrogen.59 A very low viscosity results in significant heat generation, but convection from the flowing

14 of 37

International Conference on Environmental Systems liquid is sufficient to provide cooling;57 this is of course not an option on a lunar rover, where the resulting heat would eventually evaporate the lubricant. Furthermore, motors with low viscosity lubricants function effectively only when axial loads are removed;58 although this is feasible when designing a pump, it is not as practical when designing a motor to drive a wheel. As cryogenic liquids are therefore not suitable, the remaining alternative is solid lubrication. Solid lubricants are soft or layered materials, such as lead, graphite, and molybdenum disulphide. However, graphite relies upon adsorbed moisture to provide low friction and hence is not suitable for use in vacuum.60 PTFE has also been successfully used,61 but it has low thermal conductivity; as a result, it is limited to only low-speed applications. A comprehensive list of solid lubricants potential usable for space applications can be found in Ref 60. Aside from being usable at low temperatures, other advantages of solid lubricants over liquids include reduced mass, reduced risk of decomposition in an extreme environment, greater tolerance for contaminants, reduction in time for verification tests, and no need for pressurization to prevent volatilization.60, 62 However, a major disadvantage is that solid lubricants are typically sacrificial, resulting in lower lifetimes than is achievable with liquid lubricants at more moderate temperatures. Solid lubricants have been successfully used in space applications in the past, notably in the form of molybdenum disulphide on the robotic arm and drill of the Curiosity rover.63, 64 However, the lifetime is only 27,000 revolutions, which is clearly not suitable for motors used in the mobility system. Far longer lifetimes have sometimes been achieved with solid lubricants, for example when used by SSTL to construct a reaction wheel for the GIOVE-A satellite with a lifetime of 2 billion revolutions; however, this is due to favorable and highly predictable loading. A liquid lubricant would have resulted in a lifetime many billions of revolutions higher.62 The mobility system of the Curiosity rover was initially designed with solid lubrication in mind so that it would be qualified to remain usable down to 138 K, lower than the minimum temperature of 156 K expected on Mars. However, after failing three lifetime tests, the design was switched to liquid lubrication, ultimately contributing to a launch delay of two years.56, 37 All of the Mars rovers and the ATHLETE rover use harmonic drives to obtain high gear ratios. These are alternatives to traditional gearboxes and provide very high gear ratios in a small on-axis volume. Instead of using fixed-size gears, they feature a flexible spline with teeth on the exterior of its perimeter; it is inside a fixed spline with teeth on its interior and the teeth are engaged by an elliptical wave generator rotating inside the flexible spine. There are a number of advantages in addition to the high gear ratios:65, 66

• High torque due to engaging a relatively large number of teeth on both sides of the spline. • High accuracy/repeatability, resulting from zero backlash; with conventional gears, a small amount of motion of individual gears is possible without turning any of the other gears due to small gaps between the teeth. However, the radial motion of the ﬂexible spline in a harmonic drive allows the teeth to engage very precisely with a negligible gap. • The teeth engage in a largely radial motion, eliminating much of the wear found in typical gears. As a result, harmonic drives are often rated for long lifetimes.

There is no currently available harmonic drive capable of surviving at lunar cold trap temperature with a long lifetime because of the issues mentioned above with solid lubricants. As a result, the rover presented in this work will not attempt to use solid lubrication and must maintain the actuators at temperatures above 138 K for survival and above 203 K for operation - these are the lower limits for harmonic drives rated for use on previous rovers.56, 37, 38 However, there is hope that a harmonic drive with solid lubrication will eventually be available in the form of the harmLES drive, currently being developed with funding from the EU.67

IV.D. Rough Thermal Analysis for ATHLETE IV.D.1. Wheel The selection of the wheel/tire is clearly one of the most important considerations for the ATHLETE rover, as it is the only component to be in direct contact with the lunar surface. It must remain ﬂexible and insulate the rest of the rover from signiﬁcant heat conduction into the cold surface. A wheel designed to do this already exists; although current versions of ATHLETE use standard tires while being tested in the desert on Earth, the Michelin Lunar Wheel has been developed under funding from

15 of 37

International Conference on Environmental Systems Figure 10: Michelin Lunar Wheel impacting a rock 10 cm in height. This wheel is capable of retaining its elastic properties down to at least 40 K and is made from a proprietary polymer that can insulate the rover from cold lunar surface. The wheel was designed under a NASA grant for use on the ATHLETE and CHARIOT rovers. Image reproduced from Ref 43.

NASA and can be seen in Figure 10. These wheels are based on the Michelin Tweel product, which is a non-pneumatic integrated wheel and tire made from ﬂexible polymers. The Lunar Wheel version is made from a proprietary polymer and can maintain its elastic qualities down below 40 K; in fact, the mechanical properties improve as the temperature decreases.43 Precise conductivity values are not available because of the use of a proprietary polymer, but are likely of the order of only 0.1 W/m·K, as this is a typical value of the conductivity of polymers that were described in section IV.C.2 - i.e. those suitable for use at lunar cold trap temperatures.53 There is some further consideration of thermal issues involving the wheels at the end of section IV.D.3 and section IV.D.4 below; in particular, the steady-state temperature of the wheel is shown when it is in contact with the lunar surface.

IV.D.2. Upper Body Thermal Analysis Estimates This section describes the thermal requirements for the upper body of the rover. As all components are close together in this part of the rover, this conceptual analysis simply balances heat production with emission, as effective heat distribution between closely-spaced components should not pose a significant technical challenge. As described in section IV.B, ATHLETE’s body must dissipate 3.2 kW during a traverse. It will be shown in section IV.D.3 below that 264 W will be conducted to the legs, so the remaining 2.936 kW must be emitted. The area of the upper body of the two connected Tri-ATHLETE rovers is approximately 12.4 m2, distributed as follows: Area (m2) Triangular bottom area (one Tri-ATHLETE) 2.5 Cargo platform bottom area (central section) 5 Overall area of sides 2.4 Total 12.4 These measurements were made from diagrams and photos of the Tri-ATHLETEs by comparing the rover body to certain given dimensions, such as the diameter of the wheel at 71 cm.43 Note that the area of the top of the upper body is not considered in this first order analysis, as it will be largely covered by fuel tanks and a regolith container for mined materials.

16 of 37

International Conference on Environmental Systems The emissivity of aluminum and its alloys is generally much less than 0.1, with less than 0.02 reported for some aluminum alloys at around 300 K.68 However, even an emissivity value of 0.1 is far too low to emit ∼3 kW at a reasonable operating temperature. For example, the power radiated over the given area at 300 K for this emissivity is only 570 W. Therefore an improved coating must be sought if additional radiators are to be avoided - note that the problem here is actually in keeping the rover cool, not keeping it warm. Black anodized aluminum has a thermal emissivity of 0.84,69 which is sufficient to radiate the required power at a reasonable temperature: at 267 K, 3 kW is radiated into deep space. This is lower than necessary and therefore would result in unnecessarily cold temperatures when the rover is idle, so a better approach is to only partially coat the radiative surfaces. When 7.7 m2 has been coated in this way, 3 kW is radiated at 298 K. The impact of this high emissivity coating on long-term survival at lower power levels is discussed in more detail in section IV.D.4. Having established that the rover produces more than enough heat to stay warm even without electrical heaters, at least in the main body, heating of the legs and actuators still remains to be considered. In terms of a simple power balance, the following discussion will show that the main body produces sufficient heat to keep the legs and motors warm. The total surface area of all of the legs and actuators is approximately 7 m2, estimated from images in Ref 43 and data in Ref 70. With an emissivity of only 0.1, the six legs and their motors would then emit approximately 1.9 kW at 298 K, which is a significant fraction of the heat assumed to be emitted from the main body above. This suggests that a much larger area of the body could be left with a standard aluminum finish instead of the high-emissivity black anodized coating. However, this assumes that the heat can indeed be conducted effectively into the legs. It turns out that this is not correct and so this first order attempt at a power balance is actually too simplistic; instead, a more detailed numerical analysis of thermal balance in the legs is presented in the following section IV.D.3.

IV.D.3. Actuator and Leg Thermal Analysis with a Crude Model The goal of this section is to determine if the legs of the ATHLETE rover can be adequately heated using excess heat from the main rover body, as described above in section IV.D.2. Before determining the power required for heating, it is necessary to determine the emissivity of the legs and actuators. It was indicated in the discussion of ATHLETE’s electrical power requirements in section IV.B that the joint actuators collectively require 1 kW during a typical traverse because they must compensate for small joint movements arising from the lack of suspension. If this joint power is treated as being distributed equally between all of the actuators on all of the legs, then each actuator would require 24 W. The efficiency of these actuators is around 75%, judging from data collected for the Mars Exploration Rover actuators,38 and so the power that must be dissipated from each actuator in steady-state operation would be around 7 W. However, the assumption that the power is evenly distributed is likely not always correct, as some joints are likely more susceptible to unwanted motion than others and hence require more adjustment with by the actuators. A better assumption may therefore be a maximum power dissipation of up to 21 W, which is three times the value given above; the factor of three was chosen because it allows for 2 joint actuators per leg to be active simultaneously or else all joint actuators on 2 legs. It is unlikely that any less than 2 joint actuators would be active simultaneously on one leg, as it is a complex structure and movement of any one joint would likely require compensation from at least one other. A representative surface area value for each of the actuators is 0.5 m2, estimated from images in Ref 43 and data in Ref 70. If they are coated to give an emissivity of 0.09, then the required power of up to 21 W is dissipated at a temperature of 301 K, which is below the operating 323 K limit set for actuators on the Curiosity rover.37 This first order analysis has suggested that it is possible to keep the actuators sufficiently cool during use with an appropriate emissivity choice. However, the interactions between the components from conduction have not yet been considered. Furthermore, it remains to be seen if the heat from the rover body can maintain the actuators above the minimum 138 K survival temperature (or preferably 203 K operating temperature) when idle;37 actuators may all be largely idle if the rover is temporarily parked, such as when mining, but individual actuators will also often be at least partially idle at other times - as discussed above, it is reasonable to expect that sometimes only 2 out of 6 available actuators per leg may be active. If the idle actuators can only be maintained above the survival temperature, but below the operating temperature, then some local electric heating would be required; this could be supplied simply by powering the actuator brakes and would be needed only relatively briefly prior to operation of the actuators rather than permanently throughout the idle period. Only the joint actuators have been considered up to now, but the wheel actuator must also be included.

17 of 37

International Conference on Environmental Systems Using the electrical power requirement given in section IV.B and an eﬃciency of 75%, the wheel actuator must dissipate 125 W, which can be emitted when the actuator’s surface has an emissivity of 0.5. As the ATHLETE leg is a fairly complex structure, further analytical analysis is impractical. Instead, a CAD model of a simpliﬁed ATHLETE leg was constructed so that a thermal analysis could be conducted using the software COMSOL Multiphysics.71 This analysis includes the small heat conduction through the wheel in addition to the emission described above. In the interests of simplicity for this conceptual study, one of the rotational actuators was not included and the top actuator connecting the leg to the main body was also not included. The purpose of the study is to demonstrate the feasibility of thermal control of an ATHLETE-style rover in a lunar cold trap rather than to develop a detailed design of such a thermal control system; with this mind, the parameters and some obvious additional limitations on the model are:

• All of the leg panels were assumed to be made of aluminum with emissivity 0.05 which is likely con- servatively high. However, some small components appear to contain additional unspecified materials, possibly including composites. They were nevertheless all treated as aluminum in the model. Note that the aluminum was treated as a honeycomb structure with estimated 1 mm thick walls. The overall density was estimated as 280 kg/m3 and the thermal conductivity as 14 W/m·K, using values for the interior of commercially available aluminum honeycomb72 and then accounting separately for the 1 mm walls. • The actuators were modeled as consisting primarily of steel, as many of the components are made from steel, particularly those in the harmonic drives.70 However, many other materials with lower conductivity are present too; the model partially accounted for this by lowering the density and thermal conductivity by a factor of 4 compared to COMSOL’s values for its default steel type, but this is of course only a very rough estimate. • The wheel was modeled in three sections to account for the actuator in the center, then the metal spokes seen in Figure 10, and finally the outer polymer layer. The outer section is modeled as solid PTFE, with COMSOL’s default PTFE density and thermal conductivity reduced by a factor of 10 to account for the large amounts of empty space found in the physical wheel - the empty space volume and the resulting appropriate reduction factor were estimated from various images of the wheel found in Ref 43 and other ATHLETE references mentioned in this work. The actual polymer used is unknown, as it is proprietary, and so this estimate must suffice. The emissivity is set to 0.85 to simulate a very thin coating of lunar regolith.73 A thick coating could act as additional insulation, but the wheel polymer’s conductivity is already very low and so it is unlikely that this would make a significant difference to results. The metal section is modeled as solid aluminum 6063, but with conductivity and density reduced by a factor of 20 compared to the Comsol defaults to account for the even larger empty space in this section - again, the appropriate reduction factor was estimated from empty space volume seen in wheel images. Thermal emissions from the actuator are computed using the full area seen in the photo in Figure 10, not just the limited area seen at the sides in the model in Figure 11. • All dimensions are rough estimates made by measuring images in Refs 43 and 70. • The leg’s connection to the rover body is fixed at a constant temperature of 298 K, as discussed above in section IV.D.2.

• The wheel does not move in these simulations and hence one part of the wheel is in constant contact with the ground at a ﬁxed temperature of 50 K. That part of the wheel is modeled as being partly squashed against the ground (see e.g. Figure 11), as may be expected when on rough ground or when the rover is fully loaded.

The results of two simulations will now be presented in detail. The ﬁrst was used to conﬁrm that the rover can survive when the actuators are operating at active power levels, assuming 7 W per joint actuator; the above analysis indicated that the actuators will not overheat when dissipating the assumed maximum of 21 W each, but now it must be shown that they can survive if only dissipating the average power of 7 W. Figure 11 demonstrates the result of this simulation, i.e. a rover leg in steady-state operation. The joint actuator temperature range is 238-266 K, well within the operating temperature limits of 198-323 K.37 The wheel actuator temperature of 291 K is also well within this range. Note that the thermal power conducted

18 of 37

International Conference on Environmental Systems Figure 11: Temperature profile of a rover leg in steady-state active traverse, assuming that joint actuator power is distributed as 7 W per joint. The joint actuator temperature range is 238-266 K and the wheel actuator is at 291 K. Note that the actuators in this image are the central components and the outer frame is constructed from aluminum. from the fixed-temperature rover body into the leg was found to be 44 W; i.e. the body heat assisted in keeping the leg warm by conduction. The second simulation examined the effect of a temporary idle period of 12 hours. As described later in section IV.E.2, this is much longer than expected idle periods during rover operation, as low-power mining is expected to last for only around 4 hours. This simulation result can therefore be seen as a worst case for the rover’s day-to-day idle periods. The result is shown in Figure 12. As explained in the caption, all of the actuator temperatures still remain well within the operating range, even after being idle for 12 hours. Longer idle periods are likely only during winter hibernation, which is the subject of section IV.D.4.

Figure 12: Temperature profile of a rover leg 12 hours after switching off the actuators, having previously been in the steady-state configuration seen in Figure 11. The rover body is still maintained at 298 K. All actuators are still within the survival temperature range. The joint actuator temperature range is 233-245 K and the wheel actuator is at 240 K.

In conclusion, it has been shown in this section that appropriate emissivity values for ATHLETE’s leg can be found such that all actuators remain within their survival range both when active in steady-state and even after a long idle period. Heating from the rover body assists with this, as 44 W per leg of heat input through conduction was computed in the steady-state case; this is equivalent to 264 W for all six legs

19 of 37

International Conference on Environmental Systems and this value was used earlier in section IV.D.2 to compute the remaining body heat requiring emission. Finally, note that the emissivity values presented here assumed no coating by lunar dust, but this issue will be addressed further in section IV.D.4 below.

IV.D.4. Minimum Survival Power Requirements An idle period of 12 hours was discussed for the rover legs in section IV.D.3 above, but this section will examine longer idle periods during the lunar winter, which lasts for a few days. It was found in section IV.D.2 that a high emissivity coating is required over much of the rover’s body to radiate all of the heat produced by the avionics and fuel cells when the rover is active. However, this high emissivity also means that signiﬁcant heat generation is required when the rover is idle for any signiﬁcant time; otherwise the temperature may fall to unsafe levels. This in turn means that fuel requirements for winter storage are large.

(a) (b) Figure 13: (a) An example of a set of spacecraft louvers, manufactured by Orbital Sciences. Louvers such as these can be used to reduce thermal emissivity by up to a factor of six. Image reproduced from Ref 74. (b) Demonstrating the eﬃcacy of NASA’s Lotus lunar dust-mitigation coating. Image reproduced from Ref 75.

This problem can be mitigated with the use of louvers, which effectively consist of shutters with a low-emissivity coating. When open, heat is emitted from the high-emissivity surface beneath the shutters. When closed, heat is radiated only from the shutter surface. To a rough approximation, louvers can reduce emissivity by up to a factor of six74 and an example image of a set of louvers can be seen in Figure 13a. The mass of these louvers is only 5 kg/m2 and they would cover the 7.7 m2 high-emissivity surface on the rover body described in section IV.D.2, so their impact on the rover’s payload is negligible when considering that it can carry up to 7,000 kg in lunar gravity. Unfortunately, using louvers on the lunar surface is potentially unreliable because of the possibility of being coated in high-emissivity lunar dust. However, a solution exists in the form of a new coating being developed at NASA. This coating is known as Lotus75 and consists of a microscopic layer that can be applied to metallic surfaces to prevent lunar dust from adhering. It does not affect the emissivity of the surface. The efficacy of this coating is demonstrated in Figure 13b. Not only does this coating allow for the use of louvers, but also for using some low-emissivity aluminum surfaces in the thermal model, as described in section IV.D.2 and section IV.D.3. If this coating did not exist, then it would have been necessary to assume that most of the rover would eventually become coated in dust and hence the emissivity values in the earlier thermal simulations would have needed to be increased to that of lunar regolith, i.e. to at least 0.85. Significant additional heating would then have been required. The power required for winter survival can now be calculated. The first consideration is the minimum temperature of the rover body. For simplicity, this will be set to the minimum operating temperature of the avionics, which is 243 K.76 The electronics box could be insulated to avoid needing to maintain all of the radiative surfaces at this temperature, but that would complicate cooling when the avionics are active. This temperature of 243 K will however prove to be sufficient for a demonstration of feasibility of winter survival in this conceptual analysis, especially when it is demonstrated later in this section that the power required to keep the avionics running alone is more than that required to maintain this temperature.

20 of 37

International Conference on Environmental Systems Having set this temperature, the power required to heat the legs can be demonstrated with another simulation using the simple leg geometry. As a result of the low conductivity of lunar regolith near the surface (approximately 0.011 W/m·K when using a value from Ref 77), simulation results show that there is eﬀectively no change to the surface temperature when the rover quickly moves over it or even rests in place for a few hours. However, as the rover is stationary for a number of days during the winter, its impact on the ground must now be considered. Figure 14 demonstrates the rover leg model in contact with a ground block. Rather than setting the wheel-ground contact to 50 K, as before, the lower and side edges of the ground block are now at 50 K, while the temperature of the region in contact with the wheel can change. Note however that emission from the ground and absorption by the ground are not modeled here - a model involving that is presented in section IV.D.5 below.

Figure 14: Steady-state temperature proﬁle for an idle rover leg when the rover body is at 243 K and the wheel actuator is being heated with 5 W. This represents the state of the leg in lunar winter. The side and lower boundaries of the ground block are set to 50 K, but the ground temperature in the vicinity of the wheel is free to change, as shown in section IV.D.5. The minimum actuator temperature is 140 K, compared to survival temperature of 138 K.

In steady state, which is representative of the long-term state during winter, the temperature of some of the actuators was found to fall below the minimum survival temperature of 138 K. However, this problem is solved by adding 5 W to the wheel actuator, as in Figure 14; the minimum actuator temperature is then 140 K, which is above the minimum survival temperature. At the end of the winter period, additional heating power can be applied for a short time to raise the actuator temperatures above the minimum operating temperature of 198 K. The input into the leg by conduction from the body in the simulation is 27 W. So the total power required for heating all six legs during the winter is 192 W. The additional power required for maintaining the rover body at 243 K is calculated by considering the effect of closing the louvers. When they are closed, the high-emissivity 7.7 m2 area derived in section IV.D.2 is reduced to an emissivity of 0.14. At 243 K, the total power emitted from the rover upper body is then (including both the closed louvers and the lower-emissivity surfaces) 261 W. So the total power required to maintain the rover above its survival temperature in lunar winter is 453 W. This is less than the full 500 W required for the electronics when they are fully active, but there is no need for them to be more than minimally active when the rover is idle. For example, it is possible to switch off the navigation systems and algorithms for frequently moving the leg joints to maintain a level platform. Of that 453 W, 403 W is thermal power and, as the fuel cell is 60% efficient, 181 W of that is generated directly as heat. That leaves 222 W that can potentially be used for some avionics functions, as it will then ultimately end up as heat anyway.

21 of 37

International Conference on Environmental Systems IV.D.5. Sublimation of Ice and Ground Heating Earlier sections have demonstrated that a significant amount of power is radiated from the rover, especially while it is active in traverse mode. As the objective is to mine frozen water ice from the lunar regolith, it is necessary to determine if the rover’s heat would cause some of it to sublimate before it can be mined. This section analyzes sublimation when the rover hibernates in winter, as it then remains in place for a significant time and hence heats the ground around it. Results for ground heating when the rover is active are not shown in detail because the impact on the ground temperature is negligible unless the rover remains stationary for a very long time, as described later in this section. Note that mining power is much lower than traverse power, as discussed earlier, and so the impact from mining is even smaller. The first step is to determine the temperature at which significant water ice sublimation will occur. A temperature-dependent sublimation model for ice in vacuum is presented in Ref 78. However, this model depends upon the surface area of the exposed ice at a given temperature. Around 2% of the top 1-2 m of regolith in cold traps is believed to be comprised of ice,18 as discussed in section II. The surface area is nevertheless hard to quantify, as it depends upon the ice particle size distribution. It is known that the ice does not appear in large blocks and that the maximum diameter of ice particles is 10 cm,18 but the distribution of particle sizes below this is unknown, as lunar ice measurements have so far been limited in scope. As a result of this limitation, the sublimation rate is calculated for a variety of particle sizes in Figure 15 from the 10 cm bound down to 10−5 m, which was suggested as a possible ice particle size in Ref 79. The figure can be used to obtain an indicative value for the onset of non-negligible sublimation and such a value will be helpful in interpreting the effect on sublimation of increased ground temperature. If the central line on the figure is considered to be representative (in the absence of additional information on particle size), then such an indicative value would be around 150 K. At this temperature, the sublimation rate is approximately 2 g per day per m3 of regolith; this is a measurable amount, but not significant. However, this is nevertheless a good indicative value of the onset of non-negligible sublimation because of the steep logarithmic curve on the plot - the rate dramatically increases with any temperature increase and dramatically decreases with any temperature decrease. It is worth noting that the indicated rates are only initial sublimation rates, as sublimation over time would decrease the particle sizes through mass loss; this would result in a reduction in exposed surface area and probably a corresponding reduction in the sublimation rate at later times; furthermore, as water vapor is created in the evacuated spaces between regolith grains, the vapor pressure rises to non-negligible levels and hence the net sublimation rate from the surface is considerably reduced26 compared to the rate in the model used here, which applies to a vacuum. However, this is partly compensated by a rise in temperature below the ground due to the increase thermal conductivity resulting from trapped gases instead of vacuum.78 Now that sublimation rates have been identified, the effect of the rover on ground temperature can be considered. The first part of this analysis concerns conduction through the wheel, for which a simulation was presented earlier in Figure 14. The change in ground temperature is now illustrated with a contour plot in Figure 16. It is clear then that the ground does not rise even close to the 150 K ice sublimation rate indicative value and hence there is no sublimation due to conduction through the rover’s wheel when it is idle during the winter. The second part of this analysis concerns absorption of heat emitted from the rover. As this involves a much larger heat flux into the ground, it is expected that there may be a noticeable effect over a larger area and depth. This means that a more complex model of the lunar regolith must be employed in order to capture depth-dependent conductivity and density effects. Such density and conductivity models can be found in Ref 80; they involve exponential depth-dependence and result in a dramatic temperature increase when moving below the surface, as seen in the regolith in Figure 17. In this figure, the area of absorption is equal to the approximate area of the underside of an ATHLETE rover body, so represents the relatively extreme case where all emissions are radiated directly downward and absorbed by the ground - e.g. if the rover body is close to the ground. This is therefore an illustration of the worst case for local ground heating at this power level, as all of the power is concentrated into that area. Even in this case, the high temperature does not spread away from the target area at the sides to any significant extent and so sublimated ice is confined almost entirely to the area directly below the rover; however, the depth penetration is considerably greater than for conduction, as the 150 K line now reaches down to 0.95 m. Note that, although the 1000 W value used here is greater than that found in hibernation mode, it is of

22 of 37

International Conference on Environmental Systems

104 ) 3 102

100

10−2

10−4

−6 10 1e−01 m 1e−02 m Sublimation Rate ( kg / hr m 10−8 1e−03 m 1e−04 m

−10 1e−05 m 10 130 150 170 190 210 230 Temperature ( K ) Figure 15: Initial sublimation rates of water ice per m3 of regolith at diﬀerent temperatures. The sublimation rate depends upon the exposed surface area of the ice, which in turn depends upon particle size. The ice particle size distribution is not known, but an upper bound is known to be 10 cm.18 The plot therefore indicates initial sublimation rates for all of the diﬀerent ice particle sizes shown in the legend. The image was generated using a sublimation model from Ref 78, along with the information that approximately 2% by mass of the top 1-2 m of regolith in cold traps is composed of ice.18

Figure 16: The 86 K contour in the ground for the situation described in Figure 14. It is clear then that the ground does not rise even close to the 150 K ice sublimation rate indicative value and hence there is no sublimation due to conduction through the rover’s wheel when it is idle during the winter.

23 of 37

International Conference on Environmental Systems the same order of magnitude and the above observations can still be considered representative of the impact on regolith ice in that mode. Finally, it should be emphasized again that the results presented here are for a steady-state situation; when starting from cold regolith, it takes a long time to even come close to reaching this situation, as non- negligible temperature increases are not seen beyond the top 1 mm unless the rover is stationary for many hours. Even after the first two days have passed, non-negligible temperature changes are confined to the top ∼5 cm of regolith. This is why sublimation is a concern only when parking for long-term winter hibernation. In fact, the situation shown is not reached even after the required 8 days of hibernation. However, even if it were to be reached, it is clear that sublimated ice is confined to the immediate area of the parked rover - this is of no concern, as the remaining ice in the vast area of the crater remains undisturbed.

Figure 17: 1000 W of thermal power is absorbed into a 20 m × 20 m × 10 m block of lunar regolith in steady state. The area of absorption is marked with a black line in the center and is equal to the approximate area of the underside of an ATHLETE rover body, so represents the relatively extreme case where heat is all radiated directly downward and absorbed by the ground. This is therefore an illustration of the worst case for local ground heating at this power level, as all of the power is concentrated into that area. Note that the large depth-dependent temperature variations away from the central hot area are simply due to the depth-dependent conductivity and density expressions used here; a temperature of 225 K is required at a depth of 10 m in order to give a temperature of 40 K at the surface.

IV.E. Power System Since the inside of the crater is in perpetual shadow, power cannot be generated directly with solar panels on the rovers. Beaming power from the rim to each rover was considered, but it was ultimately decided that there were too many drawbacks. Each rover would need to be tracked with very high accuracy at all times, which would require a substantial infrastructure to measure the rover positions precisely enough to hit the small receiver with the laser. Additionally, power would fluctuate as lunar dust interfered with the laser. The rover would kick up dust as it moved and excavated, which would both temporarily block part of the laser’s energy while it was in the air, and would gradually coat the receiver as it fell back down, reducing its efficiency over time. Furthermore, there are several days each year when the rim solar panels will not receive any sunlight (section V.B), during which time beaming power from the rim would be impossible. It was therefore decided that each rover should carry sufficient power onboard to operate for an entire mining sortie before returning to a central base to recharge. Based on calculations later, in section IV.E.3 and section VI.A, this can be accomplished with only 1%-2% downtime (5 minutes every 6 hours), which could potentially be reduced even further if refilling could occur in parallel with unloading the mined regolith.

IV.E.1. Fuel Cells and Batteries Since RTGs are unavailable, the main energy storage options are fuel cells and batteries. Fuel cells have a high energy density, both in terms of mass and volume, compared to batteries.81 Given the cold temperatures

24 of 37

International Conference on Environmental Systems inside the crater, the hydrogen and oxygen required to power the fuel cells should be easy to store in liquid form, which improves volumetric energy density even further. Thus, the use of fuel cells would greatly reduce the size of the fuel storage system, giving more space for the mining payload. Unfortunately, fuel cells have a much lower maximum power output than batteries.81 ATHLETE therefore uses a hybrid power system.82 The fuel cells run continuously and charge a set of high-power batteries while the rover is operating nominally. Then, whenever the rover needs to perform an energy-intensive maneuver, such as climbing, it can discharge these batteries to level the load.82 Fuel cells are powered by combining hydrogen and oxygen, which can be generated from the water that will be mined from the cold traps. When electricity is generated, the fuel cell’s water output is stored on the rover so that it can be stored and reprocessed back at base.81 This is essentially a regenerative fuel system, but with the energy and equipment for electrolysis located back at the base.

IV.E.2. Mining Analysis - a Typical Sortie As mentioned earlier, ATHLETE’s payload capacity on the Moon is 7,000 kg. If a conservative value of 1,000 kg is reserved for fuel, louvers, the shovel, regolith container, and any additional equipment, then 6,000 kg remains as the maximum regolith capacity. As explained in section IV.B, each scoop can hold 83 kg of regolith and so 72 scoops are required to ﬁll the regolith container, assuming that each scoop is full. This equates to almost four hours to load the rover when assuming that it takes one minute for loading each scoop and two minutes to lift each load into the container, as explained earlier. This time could potentially be cut in half by using two scoops - one at each end of the rover; this would not pose a problem in terms of electrical power, as the power usage during mining was shown earlier to be negligible compared to that during traverse. Assuming that only the top layer of regolith is removed on any given sortie, where the top layer is deﬁned as the layer with depth equal to the approximately 30 cm height of the scoop, a full regolith load requires the rover to mine an area of 16 m2. This is greater than the approximately 6.2 m2 of area within reach of any one of the arms (estimated from images in Ref 43), but ATHLETE can continue moving slowly while mining because, as explained in section IV.B, the mining operation requires only a very small fraction of the total available power. To identify the time required for a typical sortie, it is also necessary to know how far the rover may need to travel from its base. Consider the area within 1 km of the base, but only in a semi-circle in case one side of the base is near a crater wall that cannot be mined. The area within this 1 km semi-circle would allow for over 100,000 sorties before needing to extend the rover’s range; of course the crater surface is uneven and parts of this area may be strewn with rocks, smaller craters, or other debris, but the number of possible sorties within 1 km would remain large even after accounting for these obstructions. It is therefore safe to say that the rover need not travel more than 1 km from its base to conduct mining. If assuming that the average speed is approximately 1 km/hr (a conservative estimate, as the actual average speed is closer to 1.5 km/hr, as discussed earlier), then a sortie may last for around 6 hours in total: 2 hours for moving to and from the mining site, then another 4 hours for mining. When using the value of 2% ice in regolith by mass given earlier, this equates to 480 kg of ice mined per day.

IV.E.3. Fuel Requirements The hydrogen-oxygen fuel cell reaction is: 1 H + O → H O + energy 2 2 2 2 The Gibbs free energy, G, is the amount of energy available to do work per mole of a given substance at constant temperature and pressure. The changes in G for a given reaction can be calculated with the equation ∆G = ∆H − T ∆S, where H is enthalpy, T is constant temperature, and S is entropy for the substance. ∆H and ∆S can be obtained from a standard chemistry table83 for diﬀerent molecules, while T = 350 K is the standard operating temperature for a PEM fuel cell:84

1 H2 2 O2 H2O Change Enthalpy (kJ/mol) 0 0 −285.83 −285.83 1 Entropy (kJ/mol·K) 0.13068 2 × 0.20514 0.06991 −0.16334

25 of 37

International Conference on Environmental Systems Taking the diﬀerence between the Gibbs free energy of both sides of a chemical equation gives the amount of useful energy produced per mole of fuel consumed:

∆G = ∆H − T ∆S = −285.83 − (350 × −0.16334) = −230 kJ/mol Thus, for every mole of hydrogen (and every half-mole of oxygen) consumed by the fuel cell, 230 kJ of energy is released. If the fuel cell consumes ∆n moles of hydrogen in time ∆t, and operates the rover at an average power P and eﬃciency η, then we have:

η(∆n∆G) = P ∆t In other words, the energy released from consuming the fuel, multiplied by the efficiency of the fuel cell, is equal to the energy which is usable for electrical power in the rover. The ATHLETE rover requires approximately 5 kW of total electrical power,43 as described in section IV.B. The efficiency of the fuel cell is only ∼60%,84 which results in ∼3 kW of additional thermal energy. The equation can be rearranged to calculate the rate at which (hydrogen) fuel must be consumed to maintain this power level: ∆n P = = −130 mol/hr ∆t η∆G Conveniently, this is exactly the rate at which the electrolyzer designed to work with the first iteration of ATHLETE produces hydrogen,82 so if the base had one of these electrolyzers for each rover, it would produce just enough fuel to operate them all continuously. These results can be multiplied by each molecule’s molar mass of to determine the rate at which mass is consumed/produced by this reaction. The tanks will be designed to hold 24 hours of fuel. This allows a 200% margin on top of the 6-hour typical traverse time indicated in section IV.E.2 above, as well as allowing the rover to hibernate through the winter on its internal tanks alone, as described in section V.C. This gives the total mass of each compound that must be stored on the rover. Finally, commercially available tanks close to this size85, 86, 87 were identified and scaled to estimate the mass and volume of the tanks required to hold this much of each compound. Since tanks become more efficient as they increase in size, these numbers give an upper bound on the size of the fuel storage system. It should be possible to store all three substances in liquid form, which would minimize the volume required.

1 H2 2 O2 H2O Fuel (mol/h) −130 −65 130 Fuel (kg/h) −0.26 −2.1 2.3 Fuel (kg/24h) −6 −50 56 Tank (kg/24h) 33 40 10 Tank (L/24h) 170 50 58

The mass and volume of all three tanks must always be carried by the rover, but the mass of the fuel is transformed from one form to another, so it should be counted only on one side of the equation. This gives an estimate of the total mass and volume of the fuel storage system for the rover: 130 kg and 280 L, respectively. The current Earth-based version of ATHLETE already carries its hydrogen, so the only additional mass is from the oxygen and water, or 100 kg and 110 L. For reference from earlier, ATHLETE is 1440 kg, can carry approximately 7,000 kg of payload in lunar gravity, and is roughly 4m × 4m × 4m = 64,000 L when extended to its maximum height, so this additional mass and volume should be straightforward to add.

IV.E.4. Fuel Transfer In order to decrease turn-around time when rovers deliver to the reﬁnery unit, the electrolysis unit transfers pre-electrolyzed liquid hydrogen and oxygen to them via a fuel plumbing system. Although fuel transfer in a vacuum is signiﬁcantly more complex than in atmosphere, it has been accomplished on numerous occasions including the Robotic Refueling Mission on the International Space Station89 and regular propellant resupply of the ISS itself.90 Yarygin et al.88 studied this topic to determine the main challenges with refueling in vacuum. They found that the primary concern was surface contamination of the spacecraft with toxic fuels,

26 of 37

International Conference on Environmental Systems Figure 18: Ejection of liquid into vacuum from a 10 mm-diameter pipeline. Image reproduced from Ref 88.

which could pose problems for astronauts on EVA. Fortunately, with the use of the non-toxic resources (liquid hydrogen and oxygen) in an environment with no human presence, such concerns do not apply. However, the authors did note that, after each transfer, several litres of fuel were expelled into space during the blowdown to clear the plumbing; minimizing this loss following transfer to the mining rovers will be of utmost importance in order to maximize the exported resources.

IV.F. Navigation The ATHLETE rover’s current design does not incorporate any suitable navigation equipment for the LCT environment; it has only been tested under teleoperation in broad daylight and has no automation.44 In the lunar cold trap environment, the lack of solar illumination will confine the ways in which a rover will be able to navigate, as the ability to use standard optical cameras for imaging will be impeded. It is possible to use astronomy grade imagers to navigate. The average star light from the region of sky over the lunar south pole 91 11 2 δλ is 50 Janskies. This corresponds to 7.5 × 10 photons/s·m · λ . The albedo of the lunar highlands is 0.10- 0.15.92 If the rover has a standard luminance filter, with wavelengths from 350-700 nm, and an 10 cm × 10 cm aperture size, there will be approximately 5×108 photons per second delivered to the camera. Imaging in such low lighting is feasible with current technology, as will now be described. A commercially available Apogee Front Illuminated CCD (Charge Coupled Device) reaches a quantum efficiency of 80-90%, with a dark current of 0.05 e/p/s at -50◦C and read noise of 2 e−.93 These CCDs have roughly 15 µm pixels, which would be the equivalent of 11 photons per second per pixel without magnification, or 9 photons per second when factoring in quantum efficiency. At this sensitivity, it would be possible to take one image per second with a signal to noise ratio of 4.4. This imagning frequency is much lower than the standard 32 frames per second, but it can be overcome by either having the rover move slowly or by including other hazard avoidance imagers, described below. Sensitivity could be increased by using higher grade instruments or by increasing the number of photons falling on the CCD through the use of lenses and a larger aperture, but imaging is nevertheless possible with a common CCD and no magnification. Another available imaging technique is infrared imaging. Given black body radiation at 40 K, the peak wavelength would be 75 µm. According to the description of angular diffraction in Equation 3, the aperture of the device must be greater than 30 cm in diameter in order to have an angular resolution of 1 arcminute. 1 minute of arc corresponds to the visual acuity of a human eye with 20/20 vision.94 This makes it possible to use a decently-sized infrared imager to map out colder regions in the LCT and potential regions of interest for mining.

0.25(λ/1µm) θ = (3) D D/m In addition to the two passive imaging systems described above, the rover will also be equipped with active sensing technology. One type of active imaging that could be used is the Wide Angle Bistatic Scanning LIDAR, WABS LIDAR, developed by Neptec Design Group Ltd for the purpose of 3D imaging with rovers. Unlike regular LIDAR systems that have a minimum distance of 5 m, WABS LIDAR is functional down to 1.5 m, and has a wide ﬁeld of view (45◦ × 60◦).95 The mining operation would beneﬁt from the spacial depth of WABS LIDAR, as a large area can be mapped for hazard avoidance and distances to regions of interest can be easily calculated from LIDAR returns.

27 of 37

International Conference on Environmental Systems V. Hibernation

Because the lunar axis of rotation is inclined by 1.5◦ from a normal to the ecliptic,22 the surface is unevenly illuminated over the course of the year. In the winter, there are periods of several days when none of the solar towers will be active, as described in detail in section V.B below. At these times, the mining operation will need to be powered by diﬀerent means. To minimize the energy required in the solar-array dark periods, the mining will shut down and all power will be directed to preventing the machinery temperature from dropping below the minimum survival temperature. In the following sections, diﬀerent winter energy supply/storage systems are considered and then the total energy storage requirement is presented.

V.A. Investigated Alternatives One possible approach considered for providing energy during the winter was the use of lunar-thermal energy. Although the lack of extensive drilling on the Moon limits the availability of direct measurements of lunar temperature and conductivity at diﬀerent depths, they can nevertheless be estimated by using a speed of sound proﬁle recorded during the Apollo missions. Both heat conduction and the speed of sound are related to the density of the lunar rock; as the rock becomes denser, sound travels faster and heat conduction improves.96

Figure 19: The speed of sound at diﬀerent depths near the lunar surface. Image reproduced from Ref 96.

Unfortunately, this suggests that the base would need to drill down at least 10 m for significant thermal conductivity. Additionally, if it is necessary to drill to this depth to provide energy for the winter, then the base will become much less mobile, which will hamper mining efficiency if the area immediately surrounding the base is depleted or found to be unsuitable. Another possibility was considered in the form of taking advantage of the low temperature within the crater to use superconductors. Early superconducting materials, now known as low-temperature superconductors, demonstrated their superconducting property only at temperatures below ∼4 K. However, various high-temperature superconductors have been discovered since 1986; these materials exhibit their superconducting property up to at least 77 K, which is well above the temperature expected in a lunar cold trap during the lunar winter. The latest superconducting materials have formula REBaCuO and are known as second-generation high-temperature superconductors (HTS).97 There are two forms of superconducting energy storage: flywheels with superconducting bearings, and direct superconducting energy storage in strong magnetic fields created by currents in toroidal superconducting wires.98 The latter would theoretically last indefinitely, as currents in superconductors do not decay. However, the maximum supported current is finite, as the superconducting property of materials depends not only upon low temperature, but also the strength of the magnetic field generated by the current. At LCT temperatures, 2nd generation HTS materials can sustain a maximum magnetic field strength of approximately 5 T.99 This corresponds to a stored energy of approximately 20 MJ. Unfortunately, a strong

28 of 37

International Conference on Environmental Systems supporting structure is required in the presence of such magnetic fields and this results in a low energy density compared to the ∼15 MJ/kg offered by fuel cells.99 These systems nevertheless provide high power density, but that is not an important factor in this lunar winter storage application. Superconducting flywheels are also unsatisfactory, as they generally offer energy densities below 1 MJ/kg.100 Furthermore, although the energy takes far longer to dissipate than in a non-superconducting flywheel system, there are still losses arising from eddy currents in non-superconducting components and a significant fraction of the energy is lost over the course of a few days.101 Ultimately, the method used to store power will be the same as in other parts of the mining operation: as hydrogen and oxygen, which will be combined in fuel cells to produce energy. This choice has the advantage of using equipment and resources that will already be present at the mining site, reducing the amount of material that needs to be transported from Earth. Additionally, reusing the same technology reduces the complexity of the operation by creating fewer potential points of failure. It will become clear in section V.C below that the fuel storage requirements for winter survival are feasible when using this approach.

V.B. Days of Sunlight Solar panels on the lunar surface face two conditions under which they would be unable to produce energy: when the Sun is below the horizon due to the inclination of the lunar axis of rotation, and when the Sun is above the horizon but is blocked by a hill or mountain. Determining the optimal positions for solar panel placement to receive maximum sunlight has been the subject of prior research. Bussey et al.27 identify six potential locations, one of which is on Shackleton crater rim and would have line-of-sight with the central processing unit within the crater (point A in Figure 20). As shown in Table 1, this point receives sunlight for 81% of the year. This leaves ∼70 days per year during which the operation will be placed in hibernation mode. As can be determined from Figure 20b, the maximum period of continuous shadow on the rim solar arrays is ∼7 days.

Day 1 Day 2 Day 3 Day 4 Day 5 Day 6 Day 7 Mean for 2020 Point A 98 97 97 92 73 51 44 81

Table 1: Percentage of the time Point A is illuminated for each lunar day and for all of year 2020. Data reproduced from Ref 27.

V.C. Total Winter Power and Energy Requirements As described in section IV.D.4, almost all of the winter power requirement is supplied as heat to the rover’s main body. As a result, it is possible to achieve almost 100% efficient heat generation from whichever power source is selected during the winter, as any excess heat produced by inefficiency in electrical power production can still contribute to the total requirement. Thus, the power required to heat the rovers is almost equal to the power emitted as thermal radiation. The total requirement was found to be 453 W per rover. Winter in the crater lasts for approximately 70 days, so the total amount of energy that must be stored for each rover to survive the winter is 2.7 GJ. If this energy is provided by the electrolysis of hydrogen and oxygen in the rovers’ fuel cells, then the required mass of fuel, for each rover, is 24 kg of hydrogen and 96 kg of oxygen. However, these 70 days of winter are not continuous. By placing the solar tower at the rim location with optimal sunlight, as described in section V.B, it is possible to limit the longest continuous period of no solar energy to 7 days.27 This reduces the total fuel requirement to 2.4 kg of hydrogen and 9.6 kg of oxygen, which is less than the amount that can be stored in each rover’s onboard fuel tanks. Thus, by placing the rovers in hibernation mode, they can survive the winter without needing to refill their tanks or acquire power from some other source.

V.D. Eﬀects on Summer Power Requirement In order to stockpile fuel for the hibernation cycles, the central unit must receive more electrical power during sunlight hours than is required only for regular operation of the mining rovers. Therefore, the crater rim power station(s) must have increased generation capacity. Considering a hibernation requirement of

29 of 37

International Conference on Environmental Systems (b) (a)

(c) Figure 20: (a), (c) Six points near the lunar South Pole that receive maximum sunlight, where Point A is on the rim of Shackleton Crater. (b) Sunlight cycle for Points A and B. Images reproduced from Ref 27.

30 of 37

International Conference on Environmental Systems 453 W total power per rover during hibernation (see section IV.D.4), combined with the electrical operating power requirement of 5000 W (see section IV.B), the average power generation requirement per rover during sunlight can be calculated:

PRE DH Psunlight = + PH · (4) ηFC DS

where PRE is the average electrical power requirement of a rover during regular operation, ηFC is the eﬃciency of the rover’s fuel cell, PH is the total hibernation power requirement of a rover, DH is days per year of hibernation, and DS is days per year of sunlight. With the values given, this becomes 5000 70 P = + 453 · = 8441 W (5) sunlight 0.60 295 From this it is determined that each rover requires an average of 8441 W during periods of daylight.

VI. Energy Systems

This section details the systems that provide liquid oxygen and hydrogen fuel to the mining rovers. The subsections are listed in reverse order to allow for the inclusion of eﬃciency calculations at each step.

VI.A. Central Unit Electrolysis/Storage The primary source of energy storage in the crater will be hydrogen and oxygen, which can be combined in fuel cells to generate electricity and heat. These fuels will be generated/replenished through the electrolysis of water, powered by energy transmitted from the solar panels on the crater’s rim. Due to the low temperature in the crater, both substances can be stored in liquid form. Liquid hydrogen has a density of 70.85 kg/m3, and liquid oxygen has a density of 1140 kg/m3. This dramatically reduces the space required to store these fuel supplies, compared to compressed gas storage. Although powering an electrolysis unit within each rover (regenerative fuel cell) would be technically simpler, as the unit could be charged inductively at the base without requiring any physical connection, this option was ultimately abandoned in favor of a single large unit at the base which runs continuously and then transfers the fuel it produces to the rovers when they return to base. The primary reason for this decision was that inductively charging a regenerative fuel cell on each rover would involve substantial down-time (dependent on the fuel cell’s electrolysis capabilities) between mining sorties, greatly reducing the productivity of the mining operation, whereas a non-regenerative with fuel transfer from a central base would have rapid turn-around. Specifically, the tanks onboard the rover could be refilled in approximately 5 min through fuel transfer,86, 85 while the electrolyzer for the first iteration of ATHLETE took about 4 hours to fill its tanks.82 This electrolyzer used for the first version of ATHLETE was 20.4 kg and 0.03 m3, and the tanks only held about 82 0.5 kg of gaseous H2. If the same electrolyzer were used onboard a tri-ATHLETE, it would take about 24 hours of downtime to replenish the rover’s fuel. In order to scale up the electrolyzer to have comparable downtime to refilling the tanks, it would take 288 electrolyzer units, which would add nearly 6000 kg to the mass of a 1500 kg rover, which is clearly unreasonable. Additionally, if large tanks are constructed at the base to hold hydrogen and oxygen to refuel the rovers when they return to base, these same tanks could also be used to store fuel to heat the base during hibernation, when no power will be available from the rim-mounted solar panels. There are two main types of electrolysis in common use: Alkaline and Proton Exchange Membrane (PEM). The primary advantages of PEM electrolysis are that it can operate at significantly higher cur- 2 2 3 2 rent densities, 2.0 mA/cm vs 0.4 mA/cm , and it can produce H2 more quickly: 333 Nm /hr·m vs 190 Nm3/hr·m2 (Unit “Nm3/hr·m2” is “normalized cubic meter”, or cubic meter at standard pressure and temperature, per hour, per m2 of cell area).102 Additionally, Alkaline electrolysis requires a liquid catalyst, while PEM uses a solid catalyst, allowing PEM to operate at much higher pressures. The primary advantage of Alkaline electrolysis is a longer operating lifetime: 90,000 hours vs 20,000 hours.102 ATHLETE currently uses PEM fuel cells, but there is interest in developing an Alkaline electrolysis system that can operate effectively on the Moon.45 The reaction equation for electricity production is exactly the reverse of the fuel cell reaction:

31 of 37

International Conference on Environmental Systems 1 H O + energy → H + O 2 2 2 2 For the water electrolysis reaction, if the base dissociates ∆n moles of water, receives power P over time ∆t, and operates at eﬃciency η and temperature T , then we have:

η(P ∆t) = ∆n∆G Note that the eﬃciency factor η has switched sides from the fuel cell version of this equation. This is because electricity is now being used to generate fuel, rather than using fuel to generate electricity. This can be rearranged to determine the rate at which the electrolysis unit must receive energy from the solar panels at the rim: 1 ∆n P = ∆G η ∆t It is desirable for every rover to operate more or less continuously, which requires generating fuel at the same rate as it is burned in the rover fuel cells. Recall from section IV.E.3 that the rover produces an average of 130 mol of water with its fuel cells for every hour of operation. Thus, if the electrolysis unit is located at the base, rather than inside each rover, then at least 130 mol of water would need to be consumed by electrolysis for every hour at the base, for each rover in operation. NASA/JPL has been designing an electrolyzer for use in this type of situation, which operates at 87% eﬃciency.45 The electrolyzer uses water in its liquid state, which requires an operating temperature of at least 273.15 K. At this temperature, the kJ Gibbs free energy is approximately +240 mol . If it can be operated at the higher 350 K temperature of the kJ fuel cells, this drops to +230 mol . Thus, the base must receive power at a rate of 9.5-10 kW for each rover it operates. Note that this is only a lower bound on the operation’s power requirement, since it does not account for processing the regolith at the base.

VI.B. Wireless Power Transmission As the refinery base is situated within the region of permanent shadow and solar energy is the source of preference, power must be transferred from the rim of the crater where sunlight is accessible. Due to the long distances between the rim towers and the refinery base, as well as the difficulty of traversing the crater walls, the use of wireless power transfer has been selected for further investigation instead of a cable. The two most developed and effective methods for long-range use are microwave transfer and laser transfer. Microwave power transfer, based on early radio wave transfer by Hertz103 and Tesla,104 uses focused microwave radiation at the source in combination with a microwave-to-DC receiver.105, 106, 107 In 1974, Brown developed a device to accomplish this receiver conversion known as a “rectenna” (rectifying antenna) and successfully demonstrated its use on wirelessly powered helicopters.108 Since then, research and experimental work on microwave power transmission has continued,109 with modern experiments having demonstrated transmitter efficiencies of >80% and rectenna efficiencies of >80%.110 Microwave transmission offers a number of advantages over laser transmission, including: 1) far more resilient to poor weather conditions, 2) no risk of eye danger or spot heating, and 3) higher overall efficiency. However, it is subject to substantial beam divergence over long distances and total efficiency depends heavily on transmitter and receiver dish or phased array aperture sizes.111 Laser transmission follows a similar concept, except with the use of collimated radiation in the visible or near-infrared spectrum.112 DC power at the source is converted to optical energy using a laser diode at efficiencies up to 60.4% (as of 2004),113 and is then focused at the target using one or more optical lenses.112 At the receiver, the optical energy is converted back to DC using monochromatic (narrow-band) photovoltaic cells at an efficiency of up to 50.1% (as of 2009).114 Laser transmission is generally considered a less useful wireless power transfer technology on Earth due to the strict atmospheric constraints, but when used in a vacuum (as it would be on the lunar surface) it becomes substantially more effective. Weather interference is no longer a factor, nor are atmospheric diffusion and public safety concerns. The ability to collimate the light into an extremely tight beam offers drastic reductions in transmit/receive equipment size, as a large focusing dish or array is not required. For these reasons, wireless laser transmission was selected to power the refinery base within the crater.

32 of 37

International Conference on Environmental Systems VI.C. Solar Panel Stations The laser diodes for wireless power transmission will be placed at one or more towers on the crater rim, as detailed in section III.A. Each of these lasers will be powered by photovoltaic arrays at the tower locations, mounted on a vertical axis for rotation to follow the sun over the course of one lunar day. The average power requirement for the rim sources per continuously operational rover can be calculated as follows: PR Prim = (6) ηBE · ηLT · ηLR

where PR is the average power requirement of a rover (determined in section V.D), ηBE is the efficiency of the base electrolysis, ηLT is the efficiency of the laser power transmitter (laser diode), and ηLR is the efficiency of the laser power receiver (monochromatic photovoltaic cell). For the efficiencies detailed thus far, this becomes: 8441 W P = = 32, 341 W (7) rim 0.87 · 0.60 · 0.50 Note that this requirement per rover scales linearly with the number of rovers that operate at the base. The laser transmission tower(s) should have expansion capability for additional lasers and solar panels to accommodate growth of the mining operation. Modern photovoltaic solar cells, in a design known as “multi-junction” cells, are able to achieve an efficiency of up to 40%.115 Multi-junction cells take advantage of the high efficiencies of monochromatic cells by stacking several narrow-bandwidth cells, from shortest wavelength to longest wavelength.116 Considering that solar flux at the lunar surface (average of 1 AU) is 1,370 W/m2, a power level of approximately 548 W per square meter of solar panels can be calculated. From this and Equation 7, it can be seen that an average of 59.02 m2 of 40% efficiency solar panels is required for each continuously operating rover. In addition to this, there will be power requirements at the base for refining volatiles beyond those required for operating the rovers.

VII. Conclusion

This paper described a conceptual architecture for mining water from lunar cold traps. Specifically, it considered both the initial set-up and steady-state of a mining operation in the well-studied lunar cold trap of Shackleton crater. The environment of the crater was described, including why it is believed that minable volatiles are present, as well as the proposed overall base layout and its typical operation. Next, a more technical analysis of various aspects of this project was conducted, including rover selection, navigation, thermal issues, and power systems. The base operation, including energy generation along the rim, energy transmission to the base, and energy storage through electrolysis within the base were also discussed. Several important results have been found so far. The ATHLETE rover, which has already been developed, can easily carry the necessary payload to operate a lunar mining facility. Current technology allows for short-range vision for hazard detection in the low-light environment, and larger range position-finding using the towers should be straightforward to set up. Aside from the winter requirement of low-power heating of the wheels, the rover generates enough thermal energy to maintain itself mostly at an Earth-like temperature, despite the low ambient temperature of the crater. In fact, dissipating heat when active is a greater challenge than remaining warm during idle periods. The rover can carry fuel for at least twenty-four hours without significantly interfering with payload capacity. Solar panels positioned at locations on the crater rim for receiving maximum sunlight provide enough power to operate the base and rovers, while laser power transmission provides a means of transferring this power to the base without the infrastructure cost and complexity of laying cable down the crater slope. There is clearly much work left to be done, but preliminary results are certainly promising, and suggest that this endeavor should be achievable.

References

1Anand, M., Crawford, I., Balat-Pichelin, M., Abanades, S., Westrenen, W. v., Praudeau, G., Jaumann, R., and Seboldt, W., “A Brief Review of Chemical and Mineralogical Resources on the Moon and Likely Initial In Situ Resource Utilization (ISRU) Applications,” Planetary and Space Science, Vol. 74, 2012, pp. 42–48. 2Finetto, C., Lobascio, C., and Rapisarda, A., “Concept of a Lunar FARM: Food and Revitalization Module,” Acta Astronautica, Vol. 66, No. 9/10, 2010, pp. 1329–1340.

33 of 37

International Conference on Environmental Systems 3Belz, S., Ganzer, B., Messerschmid, E., Friedrich, K., and Schmid-Staiger, U., “Hybrid Life Support Systems with Integrated Fuel Cells and Photobioreactors for a Lunar Base,” Aerospace Science & Technology, Vol. 24, No. 1, 2013, pp. 169– 176. 4Sutton, G. P. and Biblarz, O., Rocket Propulsion Elements, John Wiley & Sons, seventh edition ed., 2001. 5United Launch Alliance, Delta IV Launch Services User’s Guide, Centennial, CO, june ed., 2013. 6Pietrobon, S. S., “Analysis of Propellant Tank Masses,” Tech. rep., Small World Communications, 2009. 7Zegler, F., Kutter, B. F., and Barr, J., “A Commercially Based Lunar Architecture,” AIAA SPACE 2009 Conference & Exposition, edited by AIAA, Pasadena, California, 2009, p. 6567. 8Kutter, B. F. and Zegler, F., “A Practical, Affordable Cryogenic Propellant Depot Based on ULA’s Flight Experience,” AIAASpace 2008 Conference & Exposition, AIAA, San Diego, California, 2008. 9Arney, D. and Wilhite, A., “Orbital Propellant Depots Enabling Lunar Architectures Without Heavy-Lift Launch Vehicles,” Journal of Spacecraft and Rockets, Vol. 47, No. 2, March 2010, pp. 353–360. 10Honour, R., Kwas, R., O’Neil, G., and Kutter, B., “Thermal Optimization and Assessment of a Long Duration Cryogenic Propellant Depot,” 50th AIAA Aerospace Sciences Meeting, AIAA, Nashville, Tennessee, 2012, p. 0630. 11Mustafi, S., Johnson, W., Kashani, A., Jurns, J., Kutter, B., Kirk, D., and Shull, J., “Subcooling for Long Duration In-Space Cryogenic Propellant Storage,” AIAA Space 2010 Conference & Exposition, AIAA, Anaheim, California, 2010, p. 8869. 12Kistler, W. P., Citron, B., and Taylor, T. C., “Lunar Commercial Mining Logistics,” AIP Conference Proceedings, Vol. 969, No. 1, 2008, pp. 260–267. 13Bienhoff, D. G., “From Importing to Exporting : The Impact of ISRU on Space Logistics,” AIAA Space 2011 Conference & Exposition, AIAA, Long Beach, California, 2011, p. 7112. 14Watson, K., Murray, B. C., and Brown, H., “The behavior of volatiles on the lunar surface,” Journal of Geophysical Research, Vol. 66, No. 9, 1961, pp. 3033–3045. 15Feldman, W. C., Lawrence, D. J., Elphic, R. C., Barraclough, B. L., Maurice, S., and Genetay, I., “Polar Hydrogen Deposits on the Moon,” Journal of Geophysical Research, Vol. 105, No. E2, 2000, pp. 4175–4195. 16Pieters, C. M., Goswami, J. N., Clark, R. N., Annadurai, M., Boardman, J., Buratti, B., Combe, J.-P., Dyar, M. D., Green, R., Head, J. W., Hibbitts, C., Hicks, M., Isaacson, P., Klima, R., Kramer, G., Kumar, S., Livo, E., Lundeen, S., Malaret, E., McCord, T., Mustard, J., Nettles, J., Petro, N., Runyon, C., Staid, M., Sunshine, J., Taylor, L. A., Tompkins, S., and Varanasi, P., “Character and Spatial Distribution of OH/H2O on the Surface of the Moon Seen by M3 on Chandrayaan-1,” Science, Vol. 326, No. 5952, 2009, pp. 568–572. 17Colaprete, A., Schultz, P., Heldmann, J., Wooden, D., Shirley, M., Ennico, K., Hermalyn, B., Marshall, W., Ricco, A., Elphic, R. C., Goldstein, D., Summy, D., Bart, G. D., Asphaug, E., Korycansky, D., Landis, D., and Sollitt, L., “Detection of Water in the LCROSS Ejecta Plume,” Science, Vol. 330, No. 6003, 2010, pp. 463–468. 18Zuber, M. T., Head, J. W., Smith, D. E., Neumann, G. A., Mazarico, E., Torrence, M. H., Aharonson, O., Tye, A. R., Fassett, C. I., Rosenburg, M. A., and Melosh, H. J., “Constraints on the Volatile Distribution Within Shackleton Crater at the Lunar South Pole,” Nature, Vol. 330, 2012, pp. 378–381. 19Nozette, S., Lichtenberg, C. L., Spudis, P., Bonner, R., Ort, W., Malaret, E., Robinson, M., and Shoemaker, E. M., “The Clementine Bistatic Radar Experiment,” Science, Vol. 274, No. 5292, 1996, pp. 1495–1498. 20Stacy, N. J. S., Campbell, D. B., and Ford, P. G., “Arecibo Radar Mapping of the Lunar Poles: A Search for Ice Deposits,” Science, Vol. 276, No. 5318, 1997, pp. 1527–1530. 21Paige, D. A., Siegler, M. A., Zhang, J. A., Hayne, P. O., Foote, E. J., Bennett, K. A., Vasavada, A. R., Greenhagen, B. T., Schofield, J. T., McCleese, D. J., Foote, M. C., DeJong, E., Bills, B. G., Hartford, W., Murray, B. C., Allen, C. C., Snook, K., Soderblom, L. A., Calcutt, S., Taylor, F. W., Bowles, N. E., Bandfield, J. L., Elphic, R., Ghent, R., Glotch, T. D., Wyatt, M. B., and Lucey, P. G., “Diviner Lunar Radiometer Observations of Cold Traps in the Moon’s South Polar Region,” Science, Vol. 330, 2010, pp. 479–482. 22Spudis, P. D., Bussey, B., Plescia, J., Josset, J.-L., and Beauvivre, S., “Geology of Shackleton Crater and the South Pole of the Moon,” Geophysical Research Letters, Vol. 35, 2008, pp. L14201. 23Berezhnoy, A., Kozlova, E., Sinitsyn, M., Shangaraev, A., and Shevchenko, V., “Origin and Stability of Lunar Polar Volatiles,” Advances in Space Research, Vol. 50, 2012, pp. 1638–1646. 24Ong, L., Asphaug, E. I., Korycansky, D., and Coker, R. F., “Volatile Retention from Cometary Impacts on the Moon,” Icarus, Vol. 207, No. 2, June 2010, pp. 578–589. 25Crider, D. H. and Vondrak, R. R., “Space Weathering Effects on Lunar Cold Trap Deposits,” Journal of Geophysical Research, Vol. 108, No. E7, 2003, pp. 5079. 26Schorghofer, N. and Taylor, G. J., “Subsurface Migration of H2O at Lunar Cold Traps,” Journal of Geophysical Research: Planets, Vol. 112, No. E2, 2007. 27Bussey, D., McGovern, J., Spudis, P., Neish, C., Noda, H., Ishihara, Y., and Srensen, S.-A., “Illumination Conditions of the South Pole of the Moon Derived Using Kaguya Topography,” Icarus, Vol. 208, No. 2, 2010, pp. 558–564. 28Rossi, C., Cunio, P. M., Alibay, F., Morrow, J., Nothnagel, S. L., Steiner, T., Han, C. J., Lanford, E., and Hoffman, J. A., “TALARIS project update: Overview of flight testing and development of a prototype planetary surface exploration hopper,” Acta Astronautica, Vol. 81, No. 1, 2012, pp. 348–357. 29Cunio, P., Babuscia, A., Bailey, J., Chaurasia, H., Goel, R., Golkar, A., Selva, D., and Timmons, E., “Initial Develop- ment of an Earth-Based Prototype for a Lunar Hopper Autonomous Exploration System,” AIAA SPACE 2009 Conference & Exposition, 2009. 30Benson, T., “Ideal Rocket Equation,” http://exploration.grc.nasa.gov/education/rocket/rktpow.html, 2013. 31Encyclopedia Astronautica, “Hydrazine,” http://www.astronautix.com/props/hydazine.htm. 32Wright, M., Kuznetsov, S., and Kloesel, K., “A Lunar Electromagnetic Launch System for In Situ Resource Utilization,” Plasma Science, IEEE Transactions on, Vol. 39, No. 1, 2011, pp. 521–528.

34 of 37

International Conference on Environmental Systems 33“Tianmen Mountains Zhangjiajie, Zhangjiajie Attractions,” http://www.visitourchina.com/guide/tianmen_ mountains.htm. 34Wirerope Works, Inc, “Specialty Strands - Ski Lift Wire Ropes,” http://www.wwwrope.com/ specialty-ropes-ski-slopes.htm. 35Fine Software, “Table of ultimate friction factors for dissimilar materials,” http://www.finesoftware.eu/ geotechnical-software/help/sheeting-design/table-of-ultimate-friction-factors-for-dissimilar-materials/, 2013. 36Engineering ToolBox, “Electrical Motor Eﬃciency,” http://www.engineeringtoolbox.com/nema-d_900.html. 37Novak, K. S., Lee, Y., and Hendricks, S., “Mars Science Laboratory Rover Actuator Thermal Design,” 40th International Conference on Environmental Systems, AIAA, 2010. 38Krishnan, S. and Voorhees, C., “The Use of Harmonic Drives on Mars Exploration Rover,” Harmonic Drive International Symposium, Nagano, Japan, 2001. 39Townsend, J., “ATHLETE Mobility Performance in Long-Range Traverse,” AIAA SPACE Conference & Exposition, AIAA, 2011, p. 7109. 40Jet Propulsion Laboratory, “The ATHLETE Rover,” http://www-robotics.jpl.nasa.gov/systems/system.cfm? System=11. 41Wilcox, B. H., “ATHLETE: Trading Complexity for Mass in Roving Vehicles,” AIAA SPACE Conference & Exposition, American Institute of Aeronautics and Astronautics, 2013. 42National Aeronautics and Space Administration, “ATHLETE Fact Sheet,” http://www.nasa.gov/pdf/390539main_ Athlete_Fact_Sheet.pdf. 43Heverly, M., Matthews, J., Frost, M., and McQuin, C., “Development of the Tri-ATHLETE Lunar Vehicle Prototype,” AMS, 2010. 44Townsend, J. and Mittman, D., “Driving ATHLETE: Analysis of Operational Eﬃciency,” AIAA SPACE Conference & Exposition, American Institute of Aeronautics and Astronautics, 2012. 45Valdez, T., “Regenerative Fuel Cells, Energy Storage Systems for Space Applications,” The von Karman Lecture Series, 2013. 46Townsend, J., Biesiadecki, J., and Collins, C., “ATHLETE Mobility Performance with Active Terrain Compliance,” 2010 IEEE Aerospace Conference, IEEE, Big Sky, Montana, 2010, p. 1530. 47Wilcox, B. H., “ATHLETE: A Cargo-Handling Vehicle for Solar System Exploration,” 2011 Aerospace Conference, IEEE, 2011, p. 1033. 48Agui, J. H., Bucek, M., Degennaro, A., Wilkinson, R. A., and Zeng, X., “Lunar Excavation Experiments in Simulant Soil Test Beds: Revisiting the Surveyor Geotechnical Data,” Journal of Aerospace Engineering, Vol. 26, 2013, pp. 117–133. 49Hurlich, A., “Low Temperature Metals,” Chemical Engineering, McGraw-Hill Publishing Co., Inc., 1963, pp. 311–325. 50Reytier, M., Kircher, F., and Levesy, B., “Characterization of titanium alloys for cryogenic applications.” AIP Conference Proceedings, Vol. 614A, DSM/DAPNIA/STCM, CEA, Centre d’Etudes Nucleaires de Saclay, Gif-sur-Yvette, France, 2002, pp. 76 – 83. 51Hickey, C. F., “Mechanical Properties of Titanium and Aluminum Alloys at Cryogenic Temperatures,” Tech. rep., Watertown Arsenal Laboratories, Watertown, MA, 1962. 52Wells, C., Whitman, T., Hannon, J., and Jensen, A., “Assembly Integration and Ambient Testing of the James Webb Space Telescope Primary Mirror,” Space 2004 Conference and Exhibit, AIAA, San Diego, California, 2004. 53Van Sciver, S. W., “Low Temperature Materials Properties,” Helium Cryogenics, chap. 2, Springer, New York, 2012, pp. 17–58. 54Johnson, T. F., Weiser, E. S., and Duong, P. G., “Cryogenic Insulation Bondline Studies for Reusable Launch Vehicles,” Tech. rep., NASA Langley, 2004. 55Weiser, E. S., Grimsley, B. W., Pipes, R. B., and Williams, M. K., “Polyimide Foams from Friable Balloons,” 47th International SAMPE Symposium and Exhibition, Long Beach, California, 2002, pp. 1151–1162. 56Novak, K., Hendricks, S., Lee, C.-J., and Orrala, C., “Mars Science Laboratory Rover Actuator Thermal Design,” Spacecraft Thermal Control Workshop 2008 , Aerospace Corporation, 2008. 57Rush, S. and Hall, L., “Tutorial on Cryogenic Submerged Electric Motor Pumps,” The 18th International Pump Users’ Symposium, Houston, Texas, 2001. 58Weisser, G. L. and Madison, J. V., “Application of Aerospace Research on Cryogenic Fuel Technology,” 34th Aerospace Sciences Meeting & Exhibit, Reno, Nevada, 1996. 59Godfrin, H., “Cryogenic Fluids,” European Advanced Cryogenics School, Chichilianne, France, 2011. 60Miyoshi, K., “Solid Lubricants and Coatings for Extreme Environments: State-of-the-Art Survey,” Tech. rep., NASA, Glenn Research Center, Cleveland, Ohio, 2007. 61McCook, N. L., Burris, D. L., Dickrell, P. L., and Sawyer, W. G., “Cryogenic Friction Behavior of PTFE based Solid Lubricant Composites,” Tribology Letters, Vol. 20, No. 2, 2005, pp. 109–113. 62Haslehurst, A. and Lewis, S., “The Use of Dry Lubricated Bearings in Reaction Wheels,” 14th European Space Mecha- nisms & Tribology Symposium ESMATS 2011 , ESA, Constance, Germany, 2011, pp. 253–261. 63Billing, R. and Fleischner, R., “Mars Science Laboratory Robotic Arm,” 12th European Space Mechanisms and Tribology Symposium, ESA, Noordwijk, The Netherlands, 2011. 64Okon, A. B., “Mars Science Laboratory Drill,” 40th Aerospace Mechanisms Symposium, NASA Kennedy Space Center, 2010. 65Schafer, I., Bourlier, P., Hantschack, F., Roberts, E. W., Lewis, S. D., Forster, D. J., and John, C., “Space Lubrication and Performance of Harmonic Drive Gears,” 11th European Space Mechanisms and Tribology Symposium, ESMATS 2005 , ESA, Lucerne, Switzerland, 2005.

35 of 37

International Conference on Environmental Systems 66Ueura, K. and Slatter, R., “Development of the Harmonic Drive Gear for Space Applications,” 8th European Symposium on Space Mechanisms and Tribology, Toulouse, France, 1999. 67Viviente, J.-l., Brizuela-parra, M., Jansson, M., and Baca, L., “HarmLES Coating Selection,” Tech. rep., EU FP7, 2012. 68Musilova, V., Hanzelka, P., Kralik, T., and Srnka, A., “Low Temperature Radiative Properties of Materials Used in Cryogenics,” Cryogenics, Vol. 45, No. 8, Aug. 2005, pp. 529–536. 69Gustavsen, A. and Berdahl, P., “Spectral Emissivity of Anodized Aluminum and the Thermal Transmittance of Alu- minum Window Frames,” Nordic Journal of Building Physics, Vol. 3, 2003. 70Harmonic Drive LLC, “Cup Type Component Sets & Housed Units, CSF & CSG Series,” http://www.harmonicdrive. net/media/support/catalogs/pdf/csf-csg-catalog.pdf. 71COMSOL Ltd, “COMSOL Multiphysics Version 4.0 Released,” Sealing Technology, Vol. 2010, No. 7, 2010, pp. 4. 72HexCel Composites, “HexWeb Honeycomb Attributes and Properties,” http://www.hexcel.com/Resources/ DataSheets/Brochure-Data-Sheets/Honeycomb_Attributes_and_Properties.pdf. 73Gaier, J. R., Street, K. W., and Gustafson, R. J., “Measurement of the Solar Absorptance and Thermal Emittance of Lunar Simulants,” 40th International Conference on Environmental Systems, AIAA, Barcelona, Spain, 2010, pp. 2010–6025. 74Orbital Sciences, “Thermal Control Louvers,” http://www.orbital.com/NewsInfo/Publications/Thermal_Louvers_ Brochure.pdf, 2004. 75Hyatt, M. J. and Straka, S. A., “The Dust Management Project: Characterizing Lunar Environments and Dust, De- veloping Regolith Mitigation Technology and Simulants,” 40th International Conference on Environmental Systems, AIAA, Barcelona, Spain, 2010, p. 6022. 76Extreme Engineering Solutions, “XCalibur1000 Datasheet,” http://www.neomore.com/XCalibur1000dsNeomore.pdf. 77Rumpf, M. E., Fagents, S. A., Crawford, I. A., and Joy, K. H., “Predicting Volatile Preservation in the Lunar Regolith Through Heat Transfer Modeling,” Lunar and Planetary Science XXXIX , League City, Texas, 2008, p. 2259. 78Andreas, E. L., “New estimates for the sublimation rate for ice on the Moon,” Icarus, Vol. 186, 2007, pp. 24–30. 79Feldman, W. C., Maurice, S., Lawrence, D. J., Little, R. C., Lawson, S. L., Gasnault, O., Wiens, R. C., Barraclough, B. L., Elphic, R. C., Prettyman, T. H., Steinberg, J. T., and Binder, A. B., “Evidence for Water Ice Near the Lunar Poles,” Journal of Geophysical Research, Vol. 106, No. E10, 2001, pp. 23231–23251. 80Vasavada, A. R., Bandﬁeld, J. L., Greenhagen, B. T., Hayne, P. O., Siegler, M. A., Williams, J.-P., and Paige, D. A., “Lunar Equatorial Surface Temperatures and Regolith Properties from the Diviner Lunar Radiometer Experiment,” Journal of Geophysical Research: Planets, Vol. 117, 2012, pp. E00H18. 81Winter, M. and Brodd, R. J., “What Are Batteries, Fuel Cells, and Supercapacitors?” Chemical Reviews, Vol. 104, No. 10, 2004, pp. 4245–4270. 82Whitacre, J., Marzwell, N., Morrison, J., Wilson, P., Mora, J., Narayanan, S., and Mathews, J., “Hybrid Power System for High Energy Density Space Systems,” International Energy Conversion Engineering Conference (IECEC), 2007. 83Schroeder, D. V., An Introduction to Thermal Physics, Addison-Wesley, 2000. 84US Department of Energy, “Comparison of Fuel Cell Technologies,” https://www1.eere.energy.gov/ hydrogenandfuelcells/fuelcells/pdfs/fc_comparison_chart.pdf, Feb. 2011. 85Directed Technologies, “Comparison of Onboard Hydrogen Storage for Fuel Cell Vehicles,” http://www. directedtechnologies.com/publications/storage/H2Storage.pdf, May 1996. 86Portable Oxygen: A User’s Perspective, “Liquid Oxygen,” http://www.portableoxygen.org/liquido2.html, 2013. 87Plastic Mart, “16 Gallon RV Waste Holding Tank,” http://www.plastic-mart.com/product/5285/ 16-gallon-rv-waste-holding-tank-h2016-wedge, 2013. 88Yarygin, V. N., Gerasimov, Y. I., Krylov, A. N., Mishina, L. V., Prikhodko, V. G., and Yarygin, I. V., “Gas Dynamics of Spacecraft and Orbital Stations (Review),” THERMOPHYSICS AND AEROMECHANICS, Vol. 18, No. 3, 2011, pp. 333–358. 89National Aeronautics and Space Administration, “Robotic Refueling Mission Factsheet,” http://ssco.gsfc.nasa.gov/ images/RRM_Factsheet.pdf. 90European Space Agency, “Europe’s Automated Ship Docks to the ISS,” http://www.esa.int/Our_Activities/Human_ Spaceflight/ATV/Europe_s_automated_ship_docks_to_the_ISS. 91Roach, F. and Megill, L., “Integrated Starlight Over the Sky,” National Bureau of Standards, 1961. 92Ulrich, G. E., Hodges, C. A., and Muehlberger, W. R., Geology of the Apollo 16 Area, National Aeronautics and Space Administration, 1981. 93Apogee Imaging Systems Inc., “High Performance Cooled CCD Camera Systems 2011,” http://www.ccd.com/pdf/ FullProductLine.pdf. 94Evans, J. M., “Standards for Visual Acuity,” ASTM International Task Group E54.08.01 On Performance Measures for Robots for Urban Search and Rescue, 2006. 95Zhu, X., DesLauriers, A., Bell, C., Gagnon, L., Guibert, M., Simard, E., Gemme, S., and Ilinca-Ignat, L., “A Wide angle Bistatic Scanning LIDAR for Navigation,” Proceedings of SPIE, Vol. 8379, 2012, p. 83790V. 96Heiken, G., Lunar Sourcebook: A User’s Guide to the Moon, CUP Archive, illustrated, reprint ed., 1991, page 93. 97Hernandez-Llambes, J. and Hazelton, D., “Advantages of Second-Generation High Temperature Superconductors for Pulsed Power Applications,” Pulsed Power Conference, 2009, pp. 221–226. 98Wen, J., Jin, J. X., Guo, Y. G., and Zhu, J. G., “Theory and Application of Superconducting Magnetic Energy Storage,” The 2006 Australasian Universities Power Engineering Conference (AUPEC’06), Melbourne, Australia, 2006. 99Selvamanickam, V., “High Temperature Superconductors for Power Applications,” Lunar Superconductivity Applications Conference, Houston, Texas, 2011. 100Chu, W.-K., “HTS Bulk Applications and Early Prototypes at TcSUH,” Lunar Superconductivity Applications Confer- ence, Houston, Texas, 2011.

36 of 37

International Conference on Environmental Systems 101Weinstein, R., Sawh, R., and Parks, D., “Trapped Field Magnets: Basics and Applications at Low Ambient Tempera- tures,” Lunar Superconductivity Applications Conference, Houston, Texas, 2011. 102Carmo, M., Fritz, D. L., Mergel, J., and Stolten, D., “A Comprehensive Review on PEM Water Electrolysis,” Interna- tional Journal of Hydrogen Energy, Vol. 38, No. 12, 2013, pp. 4901–4934. 103Susskind, C., “The Early History of Electronics. II. The Experiments of Hertz,” IEEE Spectrum, Vol. 5, No. 12, 1968, pp. 57–60. 104Hunt, I., Draper, W. W., and Brittain, J. E., “Lightning in His Hand,” IEEE Spectrum, Vol. 15, 1978, pp. 59–60. 105Takano, T., “Wireless Power Transfer From Space to Earth,” IEICE Transactions on Electronics, Vol. E96-C, No. 10, 2013, pp. 1218–1226. 106Shinohara, N., “Theory and Technologies of Wireless Power Transmission on Various Transmission Distances,” IEICE Transactions on Communications (Japanese Edition), Vol. J96-B, No. 9, 2013, pp. 881 – 893. 107Dickinson, R. M., “Power in the Sky: Requirements for Microwave Wireless Power Beamers for Powering High-Altitude Platforms,” IEEE Microwave Magazine, Vol. 14, No. 2, 2013, pp. 36–47. 108Brown, W. C., “The Technology and Application of Free-Space Power Transmission by Microwave Beam,” Proceedings of the IEEE, Vol. 62, No. 1, 1974, pp. 11–25. 109Strassner, B. and Chang, K., “Microwave Power Transmission: Historical Milestones and System Components,” Pro- ceedings of the IEEE, Vol. 101, No. 6, 2013, pp. 1379–1396. 110McSpadden, J. and Mankins, J., “Space Solar Power Programs and Microwave Wireless Power Transmission Technology,” Microwave Magazine, IEEE, Vol. 3, No. 4, 2002, pp. 46–57. 111Massa, A., Oliveri, G., Viani, F., and Rocca, P., “Array Designs for Long-Distance Wireless Power Transmission: State-of-the-Art and Innovative Solutions,” Proceedings of the IEEE, Vol. 101, No. 6, 2013, pp. 1464–1481. 112Sahai, A. and Graham, D., “Optical Wireless Power Transmission at Long Wavelengths,” 2011 International Conference on Space Optical Systems and Applications (ICSOS), 2011, pp. 164–170. 113Rossin, V., Zucker, E., Peters, M., Everett, M., and Acklin, B., “High-Power High-Eﬃciency 910- to 980-nm Broad-Area Laser Diodes,” Proceedings of SPIE, Vol. 5336, No. 1, 2004, pp. 196. 114Schubert, J., Oliva, E., and Dimroth, F., “High-Voltage GaAs Photovoltaic Laser Power Converters,” IEEE Transactions on Electron Devices, Vol. 56, No. 2, 2009, pp. 170–175. 115King, R. R., Law, D. C., Edmondson, K. M., Fetzer, C. M., Kinsey, G. S., Yoon, H., Sherif, R. A., and Karam, N. H., “40% Eﬃcient Metamorphic GaInP/GaInAs/Ge Multijunction Solar Cells,” Applied Physics Letters, Vol. 90, No. 18, 2007, pp. 183516. 116Garland, J. W., Biegala, T., Carmody, M., Gilmore, C., and Sivananthan, S., “Next-Generation Multijunction Solar Cells: The Promise of II-VI Materials,” Journal of Applied Physics, Vol. 109, No. 10, 2011, pp. 102423.

37 of 37

International Conference on Environmental Systems