A rigorous first-principles review reveals three critical errors that significantly impact model validity: the ISS solar array comparison is incorrect by 9x, space PV efficiency is underestimated by 8-10 percentage points, and the GE turbine output conflates simple-cycle with combined-cycle configurations. Most physics constants and operational parameters verify correctly, though several secondary assumptions warrant updates for improved accuracy.
ISS Solar Array Specific Power (MAJOR ERROR): The model claims 3 W/kg as the ISS comparison baseline, but verified NASA data shows 28 W/kg (31 kW ÷ 1,088 kg per Solar Array Wing). This represents a 9x error that fundamentally misrepresents the state-of-the-art for heritage space solar systems. The newer iROSA roll-out arrays installed 2021-2023 achieve >100 W/kg. This error makes Starlink's specific power improvements appear more revolutionary than they actually are relative to proven space systems.
Space PV Efficiency (SIGNIFICANT UNDERESTIMATE): The model uses 22% efficiency for space-grade solar cells, but modern triple-junction III-V cells achieve 28-32% beginning-of-life (BOL) efficiency, with end-of-life performance typically at 26-28%. Spectrolab's XTE cells deliver 32% BOL, and NREL records show 34.2% for advanced 3J cells under AM0 conditions. The 22% value is appropriate only for older silicon technology or represents excessive system losses. Correcting this assumption would significantly improve orbital system economics.
GE 7HA.03 Output Configuration Error: The model states 430 MW in combined-cycle configuration, but this figure represents simple-cycle output only. GE Vernova specifications confirm the 7HA.03 delivers 430 MW in simple cycle, scaling to 640 MW in 1×1 combined-cycle or 1,282 MW in 2×1 configuration (as deployed at Dania Beach). This distinction affects fuel consumption calculations, efficiency assumptions, and capital cost allocations.
The model's fundamental physics values align well with authoritative sources. Solar irradiance at 1361 W/m² matches NASA TSIS-1 and SORCE measurements precisely—the Total Solar Irradiance varies only ±0.1% over the 11-year solar cycle, making this an excellent engineering value. The Stefan-Boltzmann constant at 5.67×10⁻⁸ W·m⁻²·K⁻⁴ correctly rounds the exact NIST CODATA 2022 value of 5.670374419×10⁻⁸. Earth's Bond albedo of 0.30 aligns exactly with NASA CERES satellite observations.
Two minor physics discrepancies merit attention. Earth's outgoing longwave radiation should be updated from 237 to 239-240 W/m² per current NASA Earth energy budget data—a 1% difference. More notably, deep space background temperature of 3 K overestimates the CMB by approximately 10% versus the precisely measured 2.725 K (COBE/WMAP/Planck missions, Fixsen 2009). Since radiative heat transfer scales with T⁴, this approximation overestimates deep space heat rejection by ~46%, which may affect thermal equilibrium calculations depending on design margins.
Starlink satellite specifications present a nuanced picture. V1 mass at 283 kg appears approximately 9% higher than NASA NSSDCA records showing ~260 kg for the original configuration—the 283 kg figure may represent V1.5 variants with laser interlinks (260-306 kg range). V2 Mini specifications align well: the 740 kg dry mass is consistent with FCC filings stating "up to 800 kg" launch mass, and the 116 m² solar array matches Spaceflight Now's FCC-derived figure (industry sources range 105-120 m²). V3 at 1,900 kg is confirmed by multiple sources including Mobile Internet Resource Center and NextBigFuture reports from early 2025.
Power figures (7 kW, 27 kW, 60 kW) could not be verified against primary sources—SpaceX does not publicly disclose satellite power specifications in FCC filings or official documents. These appear to be derived estimates or SpaceX internal figures. The calculated specific power values (24.7-36.5 W/kg) are plausible given modern solar cell performance but should be treated as estimates rather than verified specifications.
| Version | Model | Quilty Space (May 2024) | Assessment |
|---|---|---|---|
| V1 | $230,000 | ~$200,000 | ✓ Reasonable |
| V2 Mini | $590,000 | ~$800,000 | ⚠️ 26% below consensus |
| V3 | $1.52M | ~$1.2M | ⚠️ 27% above consensus |
SpaceX does not disclose manufacturing costs, making Quilty Space's analyst estimates the most credible industry benchmark. The V1 estimate falls within the $200,000-250,000 consensus range. V2 Mini costs may be optimistic—learning curve effects and 6-satellite-per-day production rates could justify the lower figure, but $800,000 better reflects recent analyst consensus. V3 projections carry high uncertainty given pre-production status; the model's conservative $1.52M may be prudent for planning purposes despite exceeding Quilty's $1.2M projection.
The 100,000 kg payload to LEO aligns with SpaceX's official design target for reusable configuration (expendable reaches 150-200 MT). The March 2020 Starship User's Guide confirms "over 100 metric tons." Note that current hardware has demonstrated only 40-50 tons; full capability requires Block 2/3 upgrades.
Total propellant mass of ~4,600 metric tons verifies correctly, but the component breakdown requires clarification:
| Component | Model States | Verified Breakdown |
|---|---|---|
| LOX | 3,400 MT | ~3,640 MT (2,700 booster + 940 ship) |
| Methane | 1,200 MT | ~960 MT (700 booster + 260 ship) |
| Total | 4,600 MT | ~4,600 MT ✓ |
The model's stated values appear to conflate stage totals: "3,400 MT LOX" matches Super Heavy's total propellant (not just oxidizer), while "1,200 MT methane" matches Starship upper stage's total propellant. The actual LOX/methane split follows a 3.6:1 oxidizer-to-fuel mass ratio. Density conversions (LOX at 1.14 kg/L, LCH4 at 0.42 kg/L) and gallon-to-mass calculations are mathematically correct.
Heat rate at 6,200-6,370 BTU/kWh aligns with EIA AEO2025 data from Sargent & Lundy showing 6,226-6,266 BTU/kWh for H-class combined-cycle plants—this corresponds to approximately 54-55% HHV efficiency or ~64% LHV efficiency, matching GE specifications.
Natural gas BTU content should increase from 1,000 to 1,038 BTU/scf to match EIA's measured U.S. annual average. This 3.8% understatement affects fuel cost calculations proportionally.
Natural gas pricing at $4.30/MMBtu exceeds current forecasts: EIA's December 2024 Short-Term Energy Outlook projects $2.95/MMBtu average for 2025, with December 2024 spot prices at $4.61/MMBtu. The model value falls within Lazard's sensitivity range ($2.59-4.31/MMBtu) but represents a high-case scenario rather than baseline. For conservative planning, the $4.30 value may be appropriate given historical volatility ($1.51 low to >$8.50 peak in recent years).
CCGT capital cost at $1,450/kW sits between EIA overnight costs (~$868-921/kW for H-class equipment) and current market all-in costs (~$2,200-2,600/kW per Lazard LCOE v18). The model value is reasonable if including owner's costs and basic interconnection but excluding elevated market conditions that have driven recent project costs to 10-year highs.
The model's $13.80/W total ($11.60-16.00/W range) is appropriate for high-density, liquid-cooled, enterprise/hyperscale facilities. Industry benchmarks from Turner & Townsend, Dgtl Infra, and McKinsey show:
The categorical breakdown shows minor allocation differences versus industry consensus:
| Category | Model | Industry Norm | Assessment |
|---|---|---|---|
| Electrical | 38% ($5.25/W) | 40-45% | Slightly low |
| Mechanical/Cooling | 22% ($3.00/W) | 15-20% | High (justified for liquid cooling) |
| Civil & Shell | 18% ($2.50/W) | 15-21% | ✓ Within range |
| Network & Fit-out | 13% ($1.75/W) | 20-25% | Low—verify scope |
The fit-out allocation at 13% falls notably below the 20-25% industry norm, suggesting some elements may be allocated elsewhere or excluded.
PUE at 1.2 is conservative and achievable for modern liquid-cooled facilities. Google achieves 1.09 fleet-wide (1.06 best site), Meta Luleå reaches 1.07-1.08, and AWS global average is 1.15. The industry average remains 1.56, making 1.2 a reasonable engineering target.
CCGT capacity factor at 85% is optimistic compared to EIA fleet data showing 57% average (64% for newest units). However, 85% is achievable for dedicated behind-the-meter generation not subject to grid dispatch economics—the model should clarify this assumption.
GPU failure rate at 9%/year verifies precisely against Meta's Llama 3 405B training study: 220 GPU/HBM3-related failures across 16,384 H100s over 54 days yields 1.34% over that period, annualizing to ~9.1%.
Solar cell degradation at 2.5%/year is reasonable for cost-optimized LEO systems, though ISS achieves only 0.2-0.5%/year with extensive shielding. Degradation depends heavily on orbit parameters, radiation environment, and cover glass thickness.
| Parameter | Model | Verified | Status |
|---|---|---|---|
| Solar absorptivity (α) | 0.92 | 0.77-0.91 | ⚠️ High end; use 0.85-0.88 |
| Cover glass IR emissivity | 0.85 | 0.85-0.92 | ✓ Verified |
| Radiator IR emissivity | 0.90 | 0.85-0.92 | ✓ Verified |
| Max GPU Tj | 85°C | 83-90°C | ✓ Appropriate design target |
Solar cell absorptivity at 0.92 sits at the extreme high end—NASA RP-1121 shows most space cells at 0.77-0.88, with GOES cells at 0.91 being among the highest recorded.
Terminator orbit sun fraction at 98% is reasonable. Dawn-dusk sun-synchronous orbits theoretically eliminate eclipses when β=90°, though real missions like PROBA2 experience brief eclipse seasons around December. Annual average sun fractions of 95-99% are achievable.
The view factor formula (VF = 0.08 + (90 - β) × 0.002) appears to be a simplified linear approximation. The standard geometric formula for nadir-facing view factor is VF = sin²(arcsin(R_E/(R_E + h))), yielding VF ≈ 0.85 at 550 km altitude. The model's formula gives VF = 0.08-0.26 depending on β—this may represent an orbit-averaged value for non-nadir orientations but should be validated against specific geometry assumptions.
LCOE methodology follows standard NREL/DOE practice. Degradation averaging using discrete yearly summation is industry standard and introduces minimal error (<0.5% of total energy) versus continuous integration for typical 1-3%/year degradation rates. Thermal equilibrium calculations should incorporate all heat sources per NASA TFAWS guidance: direct solar, Earth albedo, planetary IR, and internal dissipation.
Critical (affects model validity):
Moderate (improves accuracy): 4. Update deep space temperature from 3 K to 2.725 K 5. Update Earth OLR from 237 to 239-240 W/m² 6. Adjust natural gas BTU from 1,000 to 1,038 BTU/scf 7. Revise Starship propellant breakdown to correct LOX/CH4 split 8. Consider updating V2 Mini cost estimate closer to $800,000 9. Document CCGT capacity factor assumption (dedicated vs. grid-dispatched)
Minor (conservative assumptions acceptable): 10. Solar absorptivity of 0.92 is high; 0.85-0.88 more typical 11. V1 Starlink mass of 283 kg is ~9% above NASA records 12. View factor formula is simplified; validate against specific geometry