Phase 3: Biological Constraints, Monte Carlo, and Mass Budget¶
Date: 2026-03-20
Model version: habitat-constraints 0.1.0 (Phase 3)
Constraints: 6 rotational + 3 biological = 9 total
Experiment script: models/habitat_constraints/experiments/run_phase3_analysis.py
Plots: models/habitat_constraints/experiments/output/
Summary¶
Phase 3 extends the model in three directions: (1) biological constraints (radiation, atmosphere, population), (2) Monte Carlo simulation to quantify uncertainty, and (3) mass budget estimation. The biological constraints do not change the feasible radius band — rotational dynamics remain dominant. But they reveal that radiation shielding is the single largest mass driver, consuming 82.8 Mt for the minimum viable cylinder. The Monte Carlo analysis shows the feasible band is robust: 96.4% of random parameter samples produce a feasible region, but the minimum radius varies widely (P5 = 286m, P95 = 3,630m), confirming that cross-coupling threshold and head turn rate are the critical unknowns.
Phase 2 → Phase 3 Comparison¶
| Metric | Phase 2 (6 constraints) | Phase 3 (9 constraints) | Notes |
|---|---|---|---|
| Constraint count | 6 rotational | 9 (+ radiation, atmo, population) | Biological added |
| Min radius at 1g | 982m | 982m | Unchanged — rotational still binding |
| Max radius at 1g | 9,177m | 9,177m | Unchanged |
| Feasible band | [982m, 9,177m] | [982m, 9,177m] | Biological constraints orthogonal to radius |
| New insight | — | Shielding = 82.8 Mt for minimum cylinder | Mass budget quantified |
| Monte Carlo feasibility | — | 96.4% of trials feasible | Band is robust |
Experiment Results¶
1. Full 9-Constraint Scorecard¶
Both reference designs pass all 9 constraints:
| Constraint | Minimum Viable (982m × 2km) | O'Neill (3,200m × 32km) |
|---|---|---|
| Vestibular (RPM) | ✓ 0.95 RPM (margin: 1.05) | ✓ 0.53 RPM (margin: 1.47) |
| Gravity level | ✓ 1.0g | ✓ 1.0g |
| Gravity gradient | ✓ 0.18% | ✓ 0.06% |
| Coriolis | ✓ ratio 0.061 | ✓ ratio 0.034 |
| Cross-coupling | ✓ 6.0 deg/s² (margin: 0.004) | ✓ 3.3 deg/s² (margin: 2.7) |
| Rim speed | ✓ 98 m/s (margin: 67%) | ✓ 177 m/s (margin: 41%) |
| Radiation shielding | ✓ 4,500 kg/m² | ✓ 4,500 kg/m² |
| Atmosphere (\(pO_2\)) | ✓ 21.3 kPa | ✓ 21.3 kPa |
| Population | ✓ 8,000 (min: 98) | ✓ 1,000,000 (min: 98) |
Key observation: The minimum viable cylinder has near-zero margin on cross-coupling (0.004 deg/s²). This is the tightest constraint in the entire system. At 982m radius, it's barely feasible — any increase in head turn rate or decrease in crew adaptation would push it infeasible.
2. Atmosphere Feasibility Map¶
The atmosphere constraint is independent of radius — it depends only on pressure and \(O_2\) fraction. The safe operating region for \(pO_2 \in [16, 50]\) kPa:
| Total Pressure | Min \(O_2\) Fraction | Max \(O_2\) Fraction | Feasible Range |
|---|---|---|---|
| 51.0 kPa (SP-413) | 31.4% | 98.0% | Wide — but 21% Earth-normal fails! |
| 56.5 kPa (NASA exploration) | 28.3% | 88.5% | Wide |
| 101.3 kPa (Earth) | 15.8% | 49.4% | Narrow — Earth-normal (21%) just inside |
Critical finding: At half-atmosphere (51 kPa), Earth-normal 21% \(O_2\) fails — \(pO_2\) = 10.7 kPa, well below the 16 kPa hypoxia threshold. The SP-413 design used 44.5% \(O_2\) at 51 kPa for this reason (\(pO_2\) = 22.7 kPa). This means you cannot simply halve the pressure without enriching the \(O_2\) percentage.
At full atmosphere (101.3 kPa), Earth-normal 21% \(O_2\) gives \(pO_2\) = 21.3 kPa — safely inside the range. But 50% \(O_2\) at full pressure already exceeds the toxicity limit (\(pO_2\) = 50.6 kPa).
Implication: Lower pressure saves structural mass but locks you into a narrower \(O_2\) management window and increases fire risk from elevated \(O_2\) percentage. For a permanent habitat, full Earth atmosphere is the safest choice.
3. Monte Carlo Simulation (500 Trials)¶
Varied 6 key parameters simultaneously with normal distributions:
| Parameter | Mean | Std Dev | Clip Range |
|---|---|---|---|
| \(\omega_{\max}\) (RPM) | 2.0 | 0.5 | [0.5, 6.0] |
| Cross-coupling threshold (deg/s²) | 6.0 | 2.0 | [2.0, 15.0] |
| Head turn rate (deg/s) | 60.0 | 15.0 | [20, 120] |
| Max Coriolis ratio | 0.25 | 0.05 | [0.05, 0.50] |
| Max rim speed (m/s) | 300.0 | 50.0 | [150, 600] |
| Max gravity gradient (%) | 1.0 | 0.3 | [0.3, 5.0] |
Results:
| Metric | Value |
|---|---|
| Feasibility rate | 96.4% |
| Min radius P5 | 286m |
| Min radius P25 | ~550m |
| Min radius P50 (median) | 971m |
| Min radius P95 | 3,630m |
| Max radius P50 | 9,274m |
Interpretation:
- The median minimum radius (971m) almost exactly matches the deterministic baseline (982m). The Monte Carlo confirms the Phase 2 result is not an outlier — it's the central tendency.
- The P5–P95 range spans more than 10× (286m to 3,630m). This enormous spread comes almost entirely from uncertainty in the cross-coupling threshold. If humans adapt better than expected (threshold > 10 deg/s²), much smaller habitats become feasible. If they don't adapt (threshold = 3 deg/s²), you need radii > 3,000m.
- 96.4% feasibility rate means only 3.6% of random parameter samples yield no feasible region at all. These are extreme cases (very low RPM tolerance + very low cross-coupling threshold simultaneously).
- The upper bound (max radius) is less variable than the lower bound, because rim speed has a tighter distribution. The median max radius of 9,274m is consistent with the deterministic 9,177m.
4. Sensitivity Analysis (Extended)¶
The top 3 most impactful parameters (±20% perturbation):
| Rank | Parameter | Radius Range | Spread |
|---|---|---|---|
| 1 | max_gravity_g |
infeasible at −20% | ∞ (breaks model) |
| 2 | max_cross_coupling_deg_s2 |
[682m, 1,533m] | 851m |
| 3 | head_turn_rate_deg_s |
[628m, 1,413m] | 785m |
max_gravity_g shows infinite spread because decreasing the maximum acceptable gravity below 0.8g while targeting 1.0g makes all designs infeasible by definition. This is not a real sensitivity — it's a constraint definition issue. The real actionable sensitivities are cross-coupling threshold and head turn rate, which together dominate the feasible band width.
All other parameters have zero spread — meaning ±20% variation in RPM tolerance, gradient tolerance, Coriolis ratio, and rim speed does not change the minimum feasible radius at 1g. These are non-binding constraints with large margins.
5. Mass Budget¶
Minimum Viable Cylinder (\(r = 982\) m, \(L = 2{,}000\) m)¶
| Component | Mass (Mt) | Fraction |
|---|---|---|
| Structural shell (steel, SF=3) | 3.3 | 3.8% |
| Radiation shielding (4,500 kg/m²) | 82.8 | 95.0% |
| Atmosphere (101.3 kPa) | 0.5 | 0.6% |
| Soil (50% area, 0.75m depth) | 0.5 | 0.5% |
| Water | 0.0 | ~0% |
| Total | ~87 Mt |
O'Neill Island Three (\(r = 3{,}200\) m, \(L = 32{,}000\) m)¶
| Component | Mass (Mt) | Fraction |
|---|---|---|
| Structural shell | 73.4 | 2.2% |
| Radiation shielding | 3,184.8 | 95.6% |
| Atmosphere | 17.2 | 0.5% |
| Soil | 52.4 | 1.6% |
| Water | 3.1 | 0.1% |
| Total | ~3,331 Mt |
The overwhelming dominance of shielding mass (95%+ in both scenarios) means that:
- Structural optimization (better materials, thinner walls) barely matters for total mass
- Atmospheric composition choices (half vs. full atmosphere) barely matter
- The only way to significantly reduce total mass is to reduce shielding requirements — either through location choice (inside the Van Allen belts), active magnetic shielding, or accepting higher radiation dose rates
This echoes the SP-413 finding from 1975: shielding dominated the Stanford torus mass budget at 9.9 Mt out of 10.5 Mt total (94%).
Visualization Outputs¶
All plots saved to models/habitat_constraints/experiments/output/:
| Plot | Description |
|---|---|
feasible_region.png |
Constraint map: radius × gravity, colored by failing constraint |
radius_sweep.png |
Pass/fail per constraint across 50–15,000m radius at 1g |
tornado_chart.png |
Sensitivity tornado showing parameter impact on min radius |
monte_carlo_histogram.png |
Distribution of min radius and feasible band width (500 trials) |
mass_budget_minimum.png |
Mass breakdown for 982m × 2km cylinder |
mass_budget_oneill.png |
Mass breakdown for 3,200m × 32km cylinder |
Key Findings¶
-
Biological constraints are orthogonal to rotational constraints. Radiation, atmosphere, and population do not change the feasible radius band — they impose requirements on shielding mass, atmospheric composition, and minimum population that are independent of cylinder geometry.
-
Radiation shielding dominates the mass budget at 95%. Everything else — structure, atmosphere, soil, water — is rounding error by comparison. This is the single most important engineering challenge.
-
The feasible band is robust under uncertainty. 96.4% of Monte Carlo trials find a feasible region. The median minimum radius (971m) matches the deterministic result (982m). The design space is not fragile.
-
But the minimum radius is highly uncertain. The P5–P95 range spans 286m to 3,630m, a 12.7× range driven almost entirely by uncertainty in cross-coupling tolerance. Resolving this uncertainty — through human centrifuge experiments at relevant rotation rates — would dramatically narrow the design space.
-
Half-atmosphere requires O₂ enrichment. You cannot simply use Earth-normal 21% O₂ at half pressure — the pO₂ drops below the hypoxia threshold. This was known in 1975 (SP-413 used 44.5% O₂) but is a critical design constraint that interacts with fire safety.
-
The minimum viable cylinder masses ~87 Mt. This is 8× the SP-413 Stanford torus (10.5 Mt) but in the same order of magnitude. The O'Neill cylinder at ~3,331 Mt is two orders of magnitude larger and requires asteroid-scale mining operations.
What This Means for the Project¶
The constraint model now has 9 constraints across 2 domains (rotational dynamics + biological). The Monte Carlo simulation confirms the results are stable. The next priorities are:
- Resolve the cross-coupling uncertainty — this is the single most valuable piece of information for habitat design
- Investigate shielding alternatives — active magnetic shielding, Van Allen belt placement (Globus and Marotta 2018), or acceptable dose rate trade-offs
- Add thermal, agricultural, and energy constraints (Phase 6) to complete the integrated model
- Build interactive visualization for parameter exploration
References¶
-
Johnson, Richard D., and Charles Holbrow, editors. Space Settlements: A Design Study. NASA SP-413, National Aeronautics and Space Administration, 1977.
-
Globus, Al, and Tom Marotta. "The High Frontier: An Easier Way." NSS Space Settlement Journal, 2018.
-
Marin, Frédéric, and Camille Beluffi. "Minimum Number of Settlers for Survival on Another Planet." Scientific Reports, vol. 10, 2020, article 9700. Nature, https://www.nature.com/articles/s41598-020-66740-0.
-
O'Neill, Gerard K. The High Frontier: Human Colonies in Space. William Morrow and Company, 1977.