Assumptions & provenance¶
Every default number in spacedc-mdao is a provenance-tagged Assumption (value, units, source, date, confidence, kind). This page is generated from the catalogs at build time. Low-confidence values are the ones to challenge first; the package keeps them as sensitivity drivers, not headline claims.
57 provenance-tagged values.
| Catalog | Entry | Field | Value | Confidence | Kind | Source |
|---|---|---|---|---|---|---|
| accelerators | amd_mi300x | mass_kg | 2 | low | estimated | Estimated OAM module + board share |
| accelerators | amd_mi300x | peak_tflops_dense | 1,300 | high | vendor | AMD MI300X: ~1.3 PFLOPS dense FP16 (2.6 with 2:4 sparsity) |
| accelerators | amd_mi300x | tdp_w | 750 | high | vendor | AMD MI300X OAM datasheet (750 W) |
| accelerators | amd_mi300x | unit_cost_usd | 1.8e+04 | low | estimated | Estimated street price, 2025-2026 |
| accelerators | generic_asic | peak_tflops_dense | 500 | low | speculative | Placeholder for a custom space-oriented ASIC |
| accelerators | generic_asic | unit_cost_usd | 1e+04 | low | speculative | Placeholder |
| accelerators | h100_sxm | ecc_mitigation | 0.8 | low | estimated | HBM ECC + scrubbing mitigates most soft errors |
| accelerators | h100_sxm | mass_kg | 1.5 | low | estimated | Estimated board+module share; SXM module mass not published |
| accelerators | h100_sxm | mem_bw_tb_s | 3.35 | high | vendor | NVIDIA H100 SXM datasheet |
| accelerators | h100_sxm | memory_gb | 80 | high | vendor | NVIDIA H100 SXM datasheet |
| accelerators | h100_sxm | peak_tflops_dense | 989 | high | vendor | NVIDIA H100 datasheet: dense FP16 tensor = 989 TFLOPS (1979 with 2:4 sparsity) |
| accelerators | h100_sxm | tdp_w | 700 | high | vendor | NVIDIA H100 datasheet (configurable TDP up to 700 W) |
| accelerators | h100_sxm | tid_tolerance_krad | 20 | low | estimated | Commercial silicon is radiation-soft; estimate before TID degradation |
| accelerators | h100_sxm | unit_cost_usd | 2.5e+04 | low | estimated | Estimated street price, 2025-2026 |
| accelerators | tpu_like | mass_kg | 1 | low | estimated | Estimated |
| accelerators | tpu_like | peak_tflops_dense | 197 | low | estimated | Order-of-magnitude bf16 for a TPU-v5e-class part |
| accelerators | tpu_like | tdp_w | 170 | low | estimated | Order-of-magnitude for a TPU-v5e-class part |
| accelerators | tpu_like | unit_cost_usd | 8,000 | low | speculative | Estimated |
| accelerators | tpu_v5e | mass_kg | 1 | low | estimated | Estimated |
| accelerators | tpu_v5e | peak_tflops_dense | 197 | high | vendor | Google TPU v5e: 197 bf16 TFLOPS per chip (published) |
| accelerators | tpu_v5e | tdp_w | 170 | low | estimated | Estimated; Google does not publish a TDP for TPU v5e |
| accelerators | tpu_v5e | unit_cost_usd | 8,000 | low | speculative | Not sold standalone (cloud only); estimated equivalent |
| batteries | li_ion_generic | cost_per_wh_usd | 0.5 | low | estimated | Estimated space-qualified pack cost |
| batteries | li_ion_generic | depth_of_discharge | 0.4 | medium | empirical | Conservative LEO cycle-life DOD |
| batteries | li_ion_generic | specific_energy_wh_per_kg | 150 | medium | empirical | Typical space-qualified Li-ion pack-level specific energy |
| batteries | li_ion_high_energy | specific_energy_wh_per_kg | 250 | low | estimated | Advanced space-qualified Li-ion (upper end) |
| chip_stacks | h100_direct_liquid | hbm_limit_k | 368 | low | estimated | HBM is the more temperature-sensitive subsystem |
| chip_stacks | h100_direct_liquid | r_junction_to_case | 0.03 | low | estimated | Direct-liquid cold-plate stack estimate for an H100-class part |
| chip_stacks | h100_direct_liquid | tj_design_k | 363 | medium | estimated | Design target ~90 C for reliability |
| chip_stacks | h100_direct_liquid | tj_max_k | 393 | high | vendor | NVIDIA H100 throttle ~120 C |
| coatings | osr | alpha_solar_bol | 0.04 | high | empirical | Spacecraft thermal heritage; OSR alpha 0.04-0.085 |
| coatings | osr | alpha_solar_eol | 0.07 | medium | estimated | 5-yr LEO degradation estimate (alpha rises) |
| coatings | osr | eps_ir_bol | 0.85 | high | empirical | OSR IR emissivity 0.85-0.87 |
| coatings | osr | eps_ir_eol | 0.82 | medium | estimated | 5-yr LEO degradation estimate (eps falls) |
| coatings | white_z93 | alpha_solar_bol | 0.14 | medium | empirical | AZ Technology / NASA; Z-93 alpha 0.13-0.18 |
| coolants | ammonia_single_phase | pump_power_fraction | 0.02 | medium | estimated | Single-phase mechanically pumped loop, optimized (NASA/ESA) |
| coolants | ammonia_two_phase | pump_power_fraction | 0.11 | medium | estimated | Hybrid two-phase loop parasitic loss ~11% (1-ACT STAIF 2007) |
| launch | aggressive_future | capacity_kg | 1e+05 | low | speculative | Speculative heavy-lift LEO capacity |
| launch | aggressive_future | cost_per_kg_usd | 200 | low | speculative | Google Suncatcher scenario: launch may fall below ~$200/kg by mid-2030s |
| launch | current_heavy_lift | capacity_kg | 1e+05 | medium | estimated | Heavy-lift reusable LEO capacity |
| launch | current_heavy_lift | cost_per_kg_usd | 1,500 | medium | estimated | Current heavy-lift reusable LEO price (mid-2020s) |
| launch | current_reusable | capacity_kg | 2.28e+04 | high | vendor | Medium-lift reusable LEO capacity |
| launch | current_reusable | cost_per_kg_usd | 3,000 | medium | estimated | Public-ish reusable LEO price, 2025-2026 |
| radiator_panels | body_mounted_osr | areal_density_kg_m2 | 2.5 | medium | empirical | NASA SoA body-mounted 2-3 kg/m^2 |
| radiator_panels | composite_deployable | areal_density_kg_m2 | 4 | low | estimated | Carbon-composite deployable panel (lighter than aluminum) |
| radiator_panels | deployable_osr | areal_density_kg_m2 | 7 | medium | estimated | Deployable system 5-10 kg/m^2 (ISS-scaled, modern mechanisms) |
| radiator_panels | high_temp_heatpipe | areal_density_kg_m2 | 3 | low | speculative | NASA lightweight high-temp radiator concept (<3 kg/m^2, 500-600 K) |
| radiators | aluminum_panel | areal_mass_kg_per_m2 | 5 | medium | estimated | Typical aluminum radiator panel areal mass incl. heat pipes |
| radiators | aluminum_panel | cost_per_m2_usd | 5,000 | low | estimated | Estimated space-qualified radiator cost |
| radiators | aluminum_panel | emissivity | 0.85 | high | empirical | Typical white-paint / OSR radiator emissivity |
| radiators | aluminum_panel | t_radiator_k | 320 | low | estimated | Panel temperature consistent with coolant loop and chip junction budget |
| radiators | aluminum_panel | t_sink_k | 250 | low | estimated | Effective LEO sink with Earth IR/albedo loading |
| solar_arrays | flexible_rosa | specific_power_w_per_kg | 150 | medium | estimated | Roll-out flexible array; higher specific power than rigid |
| solar_arrays | rigid_triple_junction | annual_degradation | 0.02 | medium | empirical | Typical LEO array degradation |
| solar_arrays | rigid_triple_junction | cell_efficiency | 0.3 | medium | empirical | Typical triple-junction cell efficiency |
| solar_arrays | rigid_triple_junction | cost_per_w_usd | 50 | low | estimated | Estimated space-qualified array cost |
| solar_arrays | rigid_triple_junction | specific_power_w_per_kg | 100 | medium | estimated | NASA SoA range ~30-200 W/kg; mid-range estimate for modern rigid arrays |