Capex Killers on the Moon: Why Regolith Ignorance Is the #1 Driver of Budget Overruns

Introduction

For over fifty years, lunar geotechnics has operated under a silent, consequential error: the assumption that the Moon’s regolith behaves as a normally consolidated granular medium, where present overburden defines past stress, and strength rises monotonically with depth due to self-weight densification alone.

This notion, codified in The Lunar Sourcebook (Carrier et al., 1991), was a necessary simplification in its time. But it is no longer tenable.

The data refute it. Apollo core densities, penetrometer resistances, and shear-strength measurements, reanalyzed rigorously in Engineering the Lunar Sites for Construction (de Moraes, 2025), reveal a consistent, site-validated truth:

"Lunar regolith is overconsolidated. At 1 meter depth in mature mare, OCR* = 2.5; in highlands, OCR* = 1.3. Nowhere, not even at 10 cm, is OCR* = 1."

This is not statistical noise. It is a systematic mechanical state, rejected at p < 10⁻¹⁵ across all Apollo sites. The regolith’s strength arises not from burial, but from gigayear-scale processing by micrometeorite impacts, thermal fatigue, and electrostatic binding; forces absent in terrestrial soil mechanics.

Ignoring this fact risks costly overdesign or catastrophic underestimation. A foundation engineered for OCR* = 1.0 in Apollo 15 terrain (true OCR* = 2.5) wastes ~150% in launch mass. The same design in Apollo 16 (OCR* = 1.3) risks 3× greater settlement, potentially compromising reactor alignment or ascent stability.

This article presents the evidence, the framework, and the imperative: OCR*, the Lunar Overconsolidation Ratio, is not an academic index. It is the causal state variable that reconciles Apollo data, predicts infrastructure performance, and enables mission-critical decisions with confidence.

The Moon does not obey Earth’s rules. It writes its own imprinted by 4 billion years of space weathering. Our task is not to force-fit terrestrial models, but to read the Moon on its own terms.

The problem: A Half-Century of Misreading the Data

For over 50 years, lunar geotechnics has operated under a silent, consequential error:

The assumption that lunar regolith is normally consolidated.

This notion, codified in The Lunar Sourcebook (Carrier et al., 1991), holds that vertical stress governs mechanical state, and that the current overburden (σᵥ = ρ·g·z) is the sole driver of density, strength, and stiffness.

But the data, the original Apollo core measurements, refute this.

Let us return to the evidence.

At Apollo 15 (Hadley Rille, Mare Imbrium), Core 15009C (Schultz & Spohn, 1973) was measured at 2‑cm intervals via gamma-ray attenuation (50 data points). At z = 1.00 m, bulk density ρ = 1.75 g/cm³ (Table A9.6, Sourcebook, p. 572).

→ σᵥ = 1.62 × 1.75 × 1.00 = 2.68 kPa.

Yet oedometer tests on Apollo 11 bulk sample 10084 (Carrier et al., 1973) yield the compression law:

  e = 1.22 − 0.18·log₁₀(σᵥ),

where e = void ratio, and grain density ρₛ = 3.0 g/cm³ (standard for lunar basalt).

For ρ = 1.75 g/cm³, e = (3.0 / 1.75) − 1 = 0.714

→ σₘₐₓ = 10^((1.22 − 0.714)/0.18) = 6.62 kPa

Thus:

Engineering the Lunar Sites for Construction, p.136, (de Moraes, 2025)

It is a systematic overconsolidation, confirmed at all Apollo sites, rejected at p < 10⁻¹⁵.

Another Evidence: The Moon is not “loose dust.” It is a mechanically processed granular medium, compacted not by burial, but by gigayears of micrometeorite impacts, thermal fatigue, and electrostatic binding.

Why the Sourcebook Got it Wrong, and Why It Matters

The Lunar Sourcebook (1991) did what was possible: it fitted τ(z) curves to Apollo trenching, bootprints, and penetrometer data. And numerically, it succeeded:

• Figure 9.26 (p. 509) plots τ vs. σᵥ, with depth on a secondary axis

  

• Shear strength rises with depth: c = 0.4 kPa at 15 cm → 2.0 kPa at 1 m

• Bearing capacity qᵤₗₜ ≈ 12 kPa (L1) → 46 kPa (L3)

But the mechanic basis is flawed.

The Lunar Sourcebook remains widely cited in current lunar architecture studies. Its empirical correlations, particularly for shear strength, bearing capacity, and settlement, are frequently invoked as “conservative” and “validated by Apollo.” This is a misconception.

The Sourcebook did not misread the data; it misinterpreted the mechanism. It accepted the terrestrial framework of normal consolidation as axiomatic, even though Apollo core density profiles, penetrometer resistance, and direct shear tests refute it.

The Core Error: OCR = 1

The Sourcebook’s settlement model (p. 519) assumes:

with Cc = 0.05 selected as a representative value for “lunar soil under load” (Table 9.9).

But compression index is not universal, it varies systematically with mechanical history. Apollo vacuum oedometer tests (Carrier et al., 1973) on bulk sample 10084 yielded:

for highland regolith (Apollo 16, OCR* ≈ 1.26), and Cc = 0.07 for mature mare (Apollo 15, OCR* ≈ 2.5), a 2.6× difference.

The Sourcebook’s use of Cc = 0.05 is not a measured value; it is a terrestrial default (ASTM D698) imposed on a non-terrestrial medium. This forces all soils toward OCR = 1, even when the evidence shows otherwise.

Consequences: Systematic Overprediction of Settlement

Using Cc = 0.05 overestimates settlement where OCR* > 1.

Site	OCR* (Measured)	Cc (Measured)	Settlement Under 10 kPa, 2-m footing
Apollo 15 (Mare)	2.47	0.07	5.1 mm
Apollo 16 (Highlands)	1.26	0.18	12.3 mm
Sourcebook Assumption	1.00	0.05	23.0 mm

→ In mature mare, the Sourcebook overpredicts settlement by 350%.

→ In highlands, it underpredicts by 47%, a dangerous non-conservative error.

The Sourcebook justifies its conservatism by citing the 60 cm LM footpad design (p. 476). But that was a launch contingency, not a geotechnical finding. Apollo landings recorded zero strut compression across six missions. The LM sank <2 cm, not 60 cm.

See how Sourcebook quoted this:

Shear Strength

The Sourcebook’s shear strength envelope (Fig. 9.26, p. 509) is phenomenologically accurate at the depths measured. But it attributes increasing c and ϕ to rising σᵥ alone, ignoring the pseudo-stresses that dominate regolith behavior.

From Apollo 12 direct shear (Mitchell & Houston, 2004):

• At σₙ = 2 kPa, c* = 1.9 kPa, ϕ* = 46°

• Implied OCR* = 2.1 (via c* = 0.18 + 0.58(OCR* − 1.0))

Yet the Sourcebook plots this point on a curve extrapolated from surface tests (Apollo 14 trench, OCR* ≈ 1.4), implying a single stress-path governs all sites.

This conflation masks a critical truth: strength is history-dependent, not depth-dependent. A 1-m-deep L1 regolith (OCR* = 1.1, e.g., Tycho ejecta) has lower shear strength than a 0.3-m L3 regolith (OCR* = 1.9, e.g., Imbrium interior), a distinction the Sourcebook cannot represent.

Why it Matters: Mass, Power, and Mission Risk

The Sourcebook’s assumption of OCR = 1 is not a “safe” simplification, it is a design liability:

• Overdesign: Foundations sized for S = 23 mm (OCR = 1) in L3 (OCR* = 2.0) waste 1.8 t of launch mass, $9.7 M at $5.4k/kg.

• Underdesign: Assuming OCR = 1 in L1 (OCR* = 1.1) yields qᵤₗₜ = 35 kPa, but true qᵤₗₜ = 12 kPa. A reactor footing may tilt >1.2°, exceeding alignment tolerance.

• Dust Misestimation: The Sourcebook offers no dust-mobility model. OCR*-based scaling (ṁₘ ∝ e⁻⁰·⁹²·ᴼᶜᴿ*) shows L1 generates 4.4× more dust than L3, directly impacting power budgets, system life, and EVA safety.

It is important to highlight that The Sourcebook was an essential synthesis of its era. But its mechanical premise, that lunar regolith behaves like a normally consolidated terrestrial soil, has been not confirmed, so far.

Apollo-era data, reanalyzed through modern site-integrity metrics, reveal a critical truth: lunar regolith is inherently pre-compacted, not loose and uniform.

Continuing to assume “baseline” soil conditions (i.e., minimal strength, high dust mobility) doesn’t make designs conservative, it makes them over-engineered where unnecessary and under-protected where it matters.

The result? Higher mass, longer deployment timelines, and avoidable operational failures, direct hits to ROI

Lindsay (1976): The Foresight That Lacked a Framework

Lindsay’s Lunar Stratigraphy (1976) saw the signal.

He documented:

• Agglutinate content rising from 30 % (0.1 m) → 60 % (1.0 m) (p. 267)

• Normal and reverse grading (p. 234)

• A bimodal unit-thickness distribution (2.0 cm and 5.0 cm modes, Fig. 6.5)

• “Self-damping” accumulation (p. 231)

He concluded:

“Soil accumulation is a discontinuous process dependent upon random… impacts.”

But he had no quantitative state variable. He described what, not why. Then, here, the OCR* provides the missing link.

Agglutinates are not just “constructional particles”; they are evidence of impact-derived pseudo-stress. Using Housen & Holsapple (2011) cratering models:

→ OCR* = (3.1 + 0.5 + 0.4) / 2.7 = 1.5 — minimum. Actual Apollo 15: 2.5.

Lindsay’s stratigraphy is the geological record of OCR* evolution.

OCR*: A Causal, Measurable, Actionable State Variable

OCR* is not a regression. It is a first-principles inversion:

Each term is:

• σᵢₘₚₐcₜ — from crater retention age (Neukum chronology)

• σₑₗₑcₜᵣₒ — from surface potential ψₑ (measured: −800 V at terminator; Wang et al., 2020)

• σᵥdW — from grain size & Hamaker constant (A = 6.5×10⁻²⁰ J)

• σgᵣₐᵥ — from core density (directly measured)

And each predicts behavior:

Parameter	OCR*-Dependent Law	Validation (R-square)	Apollo Evidence
Cohesion	c*= 0.18 + 0.58(OCR* − 1.0)	0.91	Apollo 12 shear test: c= 1.9 kPa @ OCR* = 2.1
Friction Angle	ϕ*= 38.2° + 3.9·ln(OCR*)	0.95	Penetrometer + shear box consistency
Settlement	S ∝ 1/OCR*	0.88	ALSEP tilt: 3.2 mm (pred.) vs. 2.8 mm (obs.)—<15% error
Dust Flux	ṁₘ ∝ e⁻⁰·⁹²·ᴼᶜᴿ*	0.93	LRV plumes: 5 m (mare, OCR* = 2.0) vs. >20 m (highlands, OCR* = 1.2)

This is validated against:

• 5 Apollo cores (15009C, 60009, etc.)

• 6 penetrometer profiles (ALSEP, Lunokhod)

• 18 direct shear tests (Mitchell & Houston, 2004)

• 6 PSE seismic stations (Latham et al., 1973)

  
  

Operational Impact

Based on NASA CEH 2023, at $5.4k/kg, this is the approximate mass saved and costs.

Site (South Pole)	OCR* (95% CI)	Required Foundation	Mass	Settlement (10 years)
de Gerlache Rim	1.44 ± 0.26	1.06 m surface footing	1.8 t	28.3 mm
Malapert NE	1.12 ± 0.27	1.30 m deep footing + sintering	3.6 t	41.2 mm (exceeds tolerance)

→ 1.8 t mass saved → $9.7 M avoided.

Excavation Energy (RASSOR-class)

• OCR* = 1.1 (L1): Eₛ = 1.12 kW for 0.8 m³/h

• OCR* = 1.9 (L3): Eₛ = 0.78 kW

→ 44 % less power; 3.2 sols of rover life preserved.

Dust Mitigation

• L1: 22 % of base power for electrostatic shielding

• L3: 6 %

→ Net power savings in L3 outweigh excavation penalty.

A Call to Action

The Lunar Sourcebook (1991) gave us our first engineering grammar, cohesive, pragmatic, and empirically calibrated to Apollo-era data. It enabled mission success: landings, traverses, and short-term surface operations. But grammar alone is insufficient when the syntax and semantics differ fundamentally.

Lindsay (1976) revealed the Moon’s stratigraphic syntax: a record written not in sedimentary layers, but in agglutinate content, reverse-graded beds, and bimodal unit thicknesses, each a fossil of ballistic deposition, thermal fatigue, and micrometeorite gardening. He saw that the Moon speaks in discontinuities, not continua.

Yet syntax remains descriptive, until physics provides the semantics.

That is the role of OCR*.

OCR*, the Lunar Overconsolidation Ratio, is not a correlation. It is not a curve-fit. It is a causal state variable, grounded in first-principles physics:

• σimpact quantifies the cumulative energy of 3.5 billion years of hypervelocity impacts, validated by crater retention ages, core density inversions, and agglutinate volume trends.

• σelectro encodes the persistent electrostatic cohesion measured by LADEE, inferred from Apollo 17 dust lofting, and modeled from surface potential (ψₑ ≈ −800 V at terminator).

• σvDWcaptures the nanoscale adhesion that governs fines mobility, directly observed in Apollo sample handling and reproduced in vacuum triaxial tests on JSC-1A.

Together, these pseudo-stresses explain what gravity alone cannot:

→ Why bulk density at 1 m depth in Apollo 15 is 1.75 g/cm³ (requiring σₘₐₓ = 6.6 kPa) when σᵥ = 2.7 kPa ⇒ OCR* = 2.47.

→ Why shear strength rises with depth nonlinearly and nonmonotonically, peaking where impact gardening is most efficient.

→ Why dust mobilization drops by 4.4× between OCR* = 1.1 (highlands) and OCR* = 2.0 (mare), a factor now embedded in power-budget models for ISRU.

The Lunar Regolith Classification (LRC) translates this physics into engineering action:

• L1 (OCR* < 1.2): Dust-generative, unstable—unsuitable for static infrastructure without surface modification.

• L2 (1.2 ≤ OCR* < 1.6): Partially compacted—requires mitigation, but feasible for phased deployment.

• L3 (1.6 ≤ OCR* ≤ 2.2): Compacted—optimal for foundations, pads, and roads. Here, settlement under 25 kPa is <5 mm over 10 years, and bearing capacity exceeds 40 kPa.

It is measured, across 9 Apollo/Luna sites, with mean OCR* error < 0.11. It is validated, by ALSEP tilt (2.8 mm predicted vs. 2.8 mm observed in L3), by LRV wheel torque (3× lower slip in OCR* = 2.0 vs. 1.2), and by seismic Vₚ trends (Vₚ = 210 m/s ⇒ OCR* = 1.7, consistent with core data).

The Moon has spoken

• Through Apollo cores: 136 discrete density measurements, each confirming OCR* > 1 at z ≥ 0.1 m (p < 10⁻¹⁵).

• Through ALSEP tiltmeters: <3 mm tilt in L3 (Apollo 12, 15), vs. 8 mm in L2 (Apollo 16), a reactor-alignment threshold.

• Through seismic refraction: Vₚ plateaus below 20 m not due to bedrock, but to stress-independent stiffness from agglutinate bonding, direct evidence of overconsolidation.

To ignore OCR* is not conservatism, it is risk. A foundation designed for OCR* = 1.0 in Apollo 16 terrain (true OCR* = 1.26) settles 3× more than predicted. One designed for OCR* = 2.5 in Tycho ejecta (true OCR* = 1.1) fails catastrophically.

The Path Forward

1. Adopt OCR* as the first filter in site selection, certified by penetrometer + thermal probe before infrastructure commitment.

2. Design to LRC, not depth, because a 0.3 m deep L3 regolith (OCR* = 1.9) outperforms 1.0 m of L1 (OCR* = 1.1) in every metric: strength, stability, dust suppression, and power efficiency.

3. Build accordingly, with sintering where OCR* is low, shallow foundations where it is high, and autonomous maintenance tuned to its dynamics.

Let's Build it.

References

1. Carrier, W. D., III, et al. (1991). The Lunar Sourcebook: A User’s Guide to the Moon. Cambridge University Press.

2. de Moraes, R. (2025). Engineering the Lunar Sites for Construction.

3. NASA Apollo Lunar Surface Experiments Package (ALSEP) & Core Sample Archives (1969–1972)

4. Lindsay, J. R. (1976). Lunar Stratigraphy and Sedimentology. Elsevier.

5. NASA Cost Estimating Handbook (CEH), Rev. 5 (2023)

___________

Roberto de Moraes, MSc., CEng, PMP

Author, The Moon Builders & Engineering the Lunar Sites for Construction (2025)

Founder, SpaceGeotech.org