Because the lunar surface is nearly entirely covered by unconsolidated regolith, this has been frequently invoked as a potential supply of raw materials for infrastructure, mineral resources, and fuel. When considering engineering involving the regolith, the significant differences between lunar and terrestrial soils due to differing geologic processes and overall environment should be considered. The lunar regolith particles, which are broken by bombardment, are much sharper than their terrestrial counterparts, and the agglutinates and spherical glasses, formed from impact, do not occur in the terrestrial environment. The sharper particles result in a much more abrasive material that requires special attention during the design of all equipment. In addition, the angular nature of the individual particles affects the bulk mechanics of the material, such as increasing interlocking between the angular and reentrant (locally concave) particles.
2.2.1. Bulk Properties of Lunar Samples for Construction and Operation on the Lunar Surface
 The densities of individual lunar soil grains are typically ∼3 g cm−3 (specific gravity GS = 3) with a porosity of the top 15 cm of n ∼ 0.5 and a bulk density of 1.5 g cm−3 [Mitchell et al., 1972a]. This seemingly high value for the porosity can be misleading because it includes pores within grains of complex shape. Estimates of the relative density of the lunar regolith, defined by
where e = n/(1 − n) is the void ratio, n is the porosity, emax is the maximum void ratio (least dense state) achievable, and emin is the minimum void ratio (most dense state) achievable, exceed 60% at most locations. The relative density, which is quite low very near the surface, increases significantly just 10 cm below the surface to values exceeding the maximum relative densities achievable for terrestrial soils under normal construction conditions. These high values of DR indicate that the lunar regolith is generally highly compacted. Tidal fluctuations between the Earth and Moon due to the Moon's eccentric orbit result in regular and continuous low-intensity seismic activity, which in addition to impacts of meteoroids have resulted in continuous densification of the regolith both at shallow and great depths [Carrier et al., 1991]. In microgravity experiments a lunar regolith simulant with a relative density in excess of 50% produced virtually no ejecta when impacted at speeds of less than 1 m s−1, while samples with lower relative densities produced abundant ejecta under the same conditions [Colwell and Taylor, 1999; Colwell, 2003]. An exception to the high relative densities of the lunar soil is near crater rims where the relative densities can be less than 50%, suggesting that exploration activity in these areas may result in more dust leaving the surface as well as greater penetration into the soil by astronauts, vehicles, and equipment.
 In situ soil mechanics experiments were performed on Lunakhod 1 and 2 as well as at all Apollo landing sites. Various categories of hand-held, self-recording, and ad hoc penetrometers were used on all missions as the primary soil mechanics experiment. In addition, observations of Apollo activities resulted in both qualitative and semiquantitative data. The main sources of these data include (1) astronaut observations; (2) video and still images; (3) flight mechanics telemetry; (4) bearing of objects on the lunar surface (e.g., the Lunar Module, astronauts, Early Apollo Scientific Experiments Package, and hand tools); and (5) insertion of the contingency sampler handle, the solar wind composition experiment, the flagpole, and core tubes into the surface [Sullivan, 1994]. Laboratory testing of returned samples was also performed. In situ lunar regolith has been observed to be slightly more compressible than terrestrial lunar regolith simulants, which is believed to be due to the presence of easily crushed agglutinates [Carrier et al., 1991]. The terrestrial experiments were performed on small samples at higher confining stresses than seen on the Moon. This results in lower measured shear strength because the interlocking of irregular particles that results in high strength at low pressures is either masked by high-pressure effects or does not take place at all because of the crushing of the fragile irregular agglutinates. Previous ground testing of lunar regolith has underestimated cohesion and friction angle described by the Mohr-Coulomb equation:
where τ is shear stress, c is cohesion, σ′ is effective normal stress, and ϕ is the friction angle. In addition, the load-displacement behavior of regolith is highly nonlinear, with moduli dependent on confinement level and packing density [Ko and Sture, 1980] and friction dependent on surface properties, geometric dilatancy, and loading configuration (e.g., plane, uniaxial, and triaxial). In view of these considerations, in situ measurement data are more reliable [Carrier et al., 1991]. Table 4 shows ranges for engineering properties of the in situ regolith, which to a large extent correspond to in situ regolith behavior. When densified, the strength of the regolith is quite high. This can make the regolith difficult to excavate beyond the upper ∼10 cm. This also serves to make the regolith more stable against disturbances from lunar quakes and nearby manned activity. The potential for a change in the engineering properties of processed regolith due to destruction of fragile agglutinate particles deserves attention.
Table 4. Engineering Properties of Lunar Regolitha
|Depth Range, cm||Average Bulk Density (±0.05), g cm−3||Void Ratio (±0.07)||Relative Density (±3), %||Average Cohesion, kPa||Average Friction Angle, deg|
 It is unknown whether the lunar regolith is normally consolidated or overconsolidated. Normally consolidated soils are compressed to a level in equilibrium with the load or overburden on the soil, while a soil can become overconsolidated if it is compressed by load and that load is later removed. Compressibility data, which describe the volume change, or densification, which occurs when a confining stress is applied to a soil, were obtained from ground testing of Apollo and Luna samples. For a normally consolidated regolith the compression index (Cc) ranges between 0.01 and 0.11 for dense regolith, where
where σv is the vertical stress. The recompression index, Cr (which is also given by equation (3), but the change is taken during recompression and not the initial compression), ranges between 0.000 and 0.013 [Carrier et al., 1972, 1973; Jaffe, 1973]. These values are slightly higher than basaltic lunar simulant [Carrier et al., 1972; Mitchell et al., 1974]; the irregularly shaped particles, such as the agglutinates, break easily under low confining stress and also affect the porosity of the regolith, which, in turn, affects compressibility. The coefficient of pressure at rest,
where σh is the horizontal stress and ϕ is the friction angle [Jaky, 1944], has not been measured on the Moon. Assuming the regolith is normally consolidated, it is estimated to be in the range of 0.4–0.5, comparable to terrestrial sedimentary sand, and 0.7 if recompacted against the side of a structure or large immovable object.
 Ultimate bearing capacity or strength for applied surface loading, such as footprints, footings for habitat modules, and wheel-regolith contacts, is relatively high and typically in the range of 30–1800 kPa depending on the in situ regolith density. For practical engineering purposes the limit of applied loading is controlled by allowable settlements or displacements rather than bearing capacity, except for loading applied to loose and medium-loose deposits that may exist near crater rims. On the basis of boot print analysis the allowable bearing capacity is given by [Mitchell et al., 1974; Carrier et al., 1991]
where dacc is the acceptable displacement for a 95% confidence level. For mobility purposes it is recommended that the contact pressure not exceed 1.4 kPa.
 At shallow depths (<30 cm), excavation or displacement of regolith will not pose a challenge. However, at increased depths the bulk density of the regolith increases, and with this the interlocking of particles, friction, and cohesion increase (Table 4). Excavation will become more difficult, as demonstrated during the Apollo 15 mission, when Astronaut Irwin reached a stiff layer at 30–35 cm that could not be penetrated with the scoop and required chipping to reach deeper levels [Mitchell et al., 1972b]. While terrestrial methods exist to excavate or drill in difficult conditions, they often rely on large amounts of mass and energy and the ability to replace worn equipment. These resources, with the possible exception of electrical power, cannot be assumed to be abundant for lunar exploration. Novel approaches, such as relatively lightweight vibratory equipment to loosen interlocking regolith particles, will be required for soil processing [Klosky et al., 1995, 1996; Paterson, 1992; Szabo et al., 1998].
 Recompaction following excavation may occur both from human activities as well as shakedown compaction from lunar quakes. Undisturbed regolith, below the top 30 cm, is at or above 90% relative density. Following excavation, loosely piled regolith will be at 30–40% relative density, and machine compacting can achieve on the order of 65–75% relative density [Carrier et al., 1991]. Because excavated regolith will not likely be as compact as the original undisturbed soil, designs must consider that the regolith will increase in volume by approximately 20% after excavation at depths greater than 0.5 m. In addition, it is possible that the lunar quakes will have an effect on geotechnical structures or cause settlement of structures if built on reprocessed regolith.
 A basic quantification of many mechanical properties of the lunar regolith has been performed. There are some unknowns because the material is certainly unique and the amount available for study is limited. However, there is sufficient information to begin planning and design that will meet the engineering challenges of the regolith and the lunar surface environment.
2.2.2. Lunar Regolith Simulants
 Because of the relatively limited supply of lunar soil samples and the destructive nature of standard engineering tests, lunar regolith simulants have been developed to better understand some of the bulk properties of lunar soil. The returned samples were studied to provide physical, chemical, and limited geotechnical properties, and that information was then used to select terrestrial soils that would sufficiently mimic the lunar regolith.
 During preparations for the Apollo missions, multiple lunar soil simulants were used for design and testing. Two such soils were Napa Valley Basalt (Cu = 33.0, d50 = 0.11 mm, Gs = 2.85, emax = 1.116, and emin = 0.360) [Green and Melzer, 1971] and Yuma Sand (Cu = 1.5, d50 = 0.12 mm, Gs = 2.67, emax = 0.919, and emin = 0.608) [Freitag et al., 1970], both of which were utilized for the Lunar Roving Vehicle (LRV). Although the soils were not high-fidelity simulants, they facilitated development and successful operation of the LRVs.
 When new interest in lunar exploration arose in the late 1980s, new simulants were created that built upon the knowledge gained by Surveyor, Luna, and Apollo missions. One such simulant was created by researchers at the University of Minnesota, Minnesota Lunar Simulant 1 (MLS-1), from a basaltic rock with bulk chemistry resembling Apollo 11 mare soil sample 10084. MLS-1 contains less pyroxene than the Apollo 11 lunar mares, more feldspar, a small amount (<3% by volume) of biotite, surface ferric iron (3.5% by weight) in ilmenite and mafic silicates, 0.4% water, and surface oxidation [Weiblen and Gordon, 1988; Weiblen et al., 1990]. The quarried basalt contains no glass or agglutinates, which made up the majority of sample 10084. While MLS-1 was a good mineralogical simulant, creating smaller particles and mixing soil for a representative particle-size gradation was left to individual researchers, perhaps limiting widespread use.
 MLS-1 was regraded by Perkins  to represent the range of soil distributions collected by Apollo 11, 12, 14, and 15. A series of triaxial compression experiments were performed on MLS-1 and compared to lunar regolith data, with both sets of experiments performed in Earth's 1-g atmospheric environment (Tables 5a and 5b). The stiffnesses and softening behavior were comparable, indicating graded MLS-1 closely matches the strength and stiffness properties of lunar regolith. For two confining stress levels the results for friction angle are quite close; however, when examining the cohesion terms from direct shear experiments on MLS-1 with in situ regolith, the particle-assembly cohesion (or shear strength at zero confining stress) is low. This may be due to the lack of electrostatic charging and absence of agglutinate particles, which increase interlocking behavior [Perkins, 1991]. This simulant marked an improvement over the Apollo-era simulants.
Table 5a. Lunar Regolith Simulantsa
|Percentage passing #200 sieve||43 [PM96]||36 [PM96]|
|Cu||16 [PM96]||7.5 [PM96]|
|Cc||1.1 [PM96]||1.12 [PM96]|
|d50||≈0.095 mm [PM96]||≈0.11 mm [PM96]|
|GS||3.2 [M94]||2.91 [W95]|
|emax/ρmin, g cm−3||1.05/1.56 [PM96]||1.18/1.33 [PM96] —/1.43 [K2000]|
|emin/ρmax, g cm−3||0.45/2.20 [PM96]||0.61/1.80 [PM96] —/1.83 [K2000]|
|Aspect ratio|| ||0.68 [W95]|
|Shear strength||ϕ = 58° for c = 0 at p = 10 kPa [PM96]||ϕ = 64° for c = 0 at p = 10 kPa [PM96]|
| || ||ϕ = 45°; c ≤ 1 kPa at ρ = 1.5–1.65 g cm−3 [M94]|
| || ||ϕ = 49°; c = 0.2 kPa at ρ = 1.9 g cm−3 [P91]|
| || ||ϕ = 52–55°; c = 2.4–3.8 kPa [C91]|
| || ||ϕ = 44.4°; c = 3.9 kPa at ρ = 1.62 g cm−3 [K96]|
| || ||ϕ = 52.7°; c = 13.4 kPa at ρ = 1.72 g cm−3 [K96]|
|Dilatancy angle|| ||44.0° at p = 1 kPa and ρ = 1.62 g cm−3 [K96]|
| || ||40.5° at p = 10 kPa and ρ = 1.62 g cm−3 [K96]|
| || ||65.0° at p = 10 kPa and ρ = 1.72 g cm−3 [K96]|
|Residual strength||44° [PM96]||42° [PM96]|
|Young's modulus E, MPa||4.60 at DR = 37% [P91] 7.99 at DR = 66% [P91] 7.92 at DR = 97% [P91]||18–60 at DR = 40% [K2000] 65–110 at DR = 60% [K2000] (higher values are at higher stresses)|
|Bulk modulus K, MPa||DR = 37% K = 9.63 MPa [P91] DR = 66% K = 7.69 MPa [P91] DR = 97% K = 12.1 MPa [P91]||DR = 40%, K = 35–60 [K2000] DR = 60%, K = 75–110 [K2000] (higher values are at higher stresses)|
Table 5b. Lunar Regolith Samples
|Percentage passing #200 sieve||52 [C03]|
|d50||0.072 mm [C03]|
|GS||2.9–3.2, 3.1 recommended [C91]|
|emax/ρmin (g cm−3)||1.39/1.26 (Apollo 11 [Cr70])|
| ||—/1.15 (Apollo 12 [J71])|
| ||2.26–2.37/0.87–0.89 (Apollo 14 [C73])|
| ||1.94/1.10 (Apollo 15 [C73])|
|emin/ρmin (g cm−3)||0.67/1.80 (Apollo 11 [Cr70])|
| ||—/1.93 (Apollo 12 [J71])|
| ||0.87–0.94/1.55–1.51 (Apollo 14 [C73])|
| ||0.71/— (Apollo 15 [C73])|
|Elongation||1.31–1.39 (Mahmood et al., unpublished report, 1974)|
|Aspect ratio||0.4–0.7 [Görz et al., 1972]|
|Shear strength||ϕ = 30–50°; c = 0.1–1.0 kPa [Mitchell et al., 1974]|
| ||ϕ = 42°; c = 0.52 kPa at 0–15 cm depth [C91] ϕ = 46°; c = 0.90 kPa at 0–30 cm depth [C91] ϕ = 54°; c = 3.0 kPa at 30–60 cm depth [C91] ϕ = 49°; c = 1.6 kPa at 0–60 cm depth [C91]|
 Prompted by the Space Exploration Initiative in 1989, a group of researchers convened a workshop to define requirements for a standard simulant to be used by the research community. The result was Johnson Space Center 1 (JSC-1), a low-titanium mare-like soil with a high percentage of glass. The soil was taken from a volcanic ash deposit near Flagstaff, Arizona, and sieved, and larger particles were crushed in an impact mill. The soil contained mainly plagioclase, pyroxene, and olivine but also a high percentage of (nonspherical) glass more closely matching a low-titanium mare soil [Willman et al., 1995]. Large quantities of JSC-1 were produced, allowing distribution to several researchers. Major elements of JSC-1 include silicon oxide, aluminum oxide, and calcium oxide, making it comparable to MLS-1 and Apollo 14 lunar sample 14163 [McKay et al., 1994]. Like MLS-1 it does not contain agglutinates that would increase material nonlinearities [Perkins, 1991]. JSC-1 is presently the industry standard lunar regolith simulant distributed widely for research and education.
 Willman et al.  performed conventional triaxial tests on JSC-1 at 20.6, 34.4, and 68.7 kPa, equivalent to regolith at depths of 8–26 m. Specimens were prepared by tamping in three lifts, at densities of 1.50, 1.60, and 1.65 g cm−3, ranging from 20 to 60% relative density. A friction angle of 45° and cohesion of 1.0 kPa were reported for all densities, which is an unusual result: Strength is expected to vary with density as well as confining pressure.
 Klosky et al. [1996, 2000] performed multiple triaxial tests of JSC-1 at pressures ranging from 1 to 80 kPa with samples prepared using base vibration at three density levels (1.62, 1.71, and 1.81 g cm−3) corresponding to 53, 75, and 95% relative density (Tables 5a and 5b). The material shows high friction angles and apparent cohesion, as does lunar regolith. In addition to verifying that the mechanical characteristics of the material were an appropriate simulation of lunar regolith, Klosky et al.  also demonstrated the importance of the environment in which the material exists. The increasing trend in dilatancy angle with decreasing confining stress supports the Klosky et al. observation that the Mohr-Coulomb model used to determine friction angle and cohesion showed nonlinearity at low pressures. That is, constitutive behavior, or the way the soil contracts and expands under loading and unloading, is dependent on the confining stress on the material, particularly at low confining stress. With the reduced gravity on the Moon it is therefore particularly important that the confining stress be considered when determining load conditions for construction on and in lunar regolith.
 Currently, little of the MLS-1 or JSC-1 simulants remain in distribution, however, and limited quantities of other simulants, such as the Japanese FJS-1 and MKS-1, are available. To provide a large quantity for wide distribution, production of a new lunar soil simulant in the United States is underway with the support of NASA. Internationally, several other businesses and organizations are also producing simulants. Simulants offer the opportunity to allow extensive experimental testing as the quantity available is much greater than lunar regolith. It can be used to provide test beds for various exploration activities and equipment designs as well as model verification. While an important asset, the ability to simulate the lunar environment should be evaluated carefully. Simulants do differ from the lunar regolith (particle shapes, mineralogical composition, size distribution to varying degrees), and it is important to be aware of those differences in relation to how the simulant is to be used. For instance, for geotechnical engineering the detailed chemical composition is not as important as the grain shape or size distribution. For chemical processing, though, the composition may be more important. The goodness of a simulant has to be examined on a case-by-case basis. Also, the simulant (and lunar regolith) behavior is dependent on temperature, gravity, vacuum, etc., conditions of the environment in which it is used; to be most effective, the environment should be simulated as closely as possible. Where that is not feasible, the limitations of simulating the lunar environment on Earth should be taken into account when extrapolating to the Moon.