Methodology: 05_trade_study_methodology.md. Source data: ../analysis/results/variantMetrics.csv, ../analysis/results/tradeScores.csv, ../analysis/results/mcWinShare.csv, produced by ../analysis/pipeline/runVariantAnalysis.m and ../analysis/pipeline/runTradeStudy.m. Variant design concepts are described in 04_physical_variants.md.
The trade study has been run. This document reports the full results and the recommendation that follows from them.
Live pipeline note: the figures under docs/figures now track the full-comparison rerun over all three variants. If the formal gate narrows the candidate set, the compliant-only decision rerun is published separately under ../analysis/results/*_compliant.* and figures/compliant/ so these comparison figures are not overwritten.
Note: the embedded figure files track the live analysis pipeline (regenerated on every full run — currently behavioral-era numbers in the dark house theme), while the tables in this document record the static-era values as of the original study. Where they differ, the tables are the as-of-then record.
| Metric | HyperCook (A) | LeanBroth (B) | EverSimmer (C) | SR cap |
|---|---|---|---|---|
| Mass, kg (utilization) | 14,320 (95.5%) | 7,570 (50.5%) | 11,120 (74.1%) | ≤15,000 (SR-GS-011) |
| Power, kW (utilization) | 498 (99.6%) | 239 (47.8%) | 363 (72.6%) | ≤500 (SR-GS-012) |
| Cost, kCr (utilization) | 1,980 (99.0%) | 1,070 (53.5%) | 1,905 (95.3%) | ≤2,000 (SR-GS-013) |
| Volume, m³ (utilization) | 397 (99.3%) | 240 (60.0%) | 297 (74.3%) | ≤400 (SR-GS-014) |
| Throughput, bph (margin) | 320 (+60%) | 210 (+5%) | 240 (+20%) | ≥200 (SR-GS-002) |
| Automation, avg | 0.944 | 0.800 (zero margin) | 0.956 | ≥0.8 (SR-GS-003) |
| Operators | 3.8 | 4.9 | 2.7 | ≤5 (SR-GS-004) |
| Availability | 0.9591 | 0.9720 | 0.9789 | — (informative) |
| N-1 capacity retention | 0% | 0% | 66.7% (160 bph after losing one cell) | — (informative) |
| Leaf component count | 19 | 16 | 25 (3 nested ProductionCells x 5 units + 10 top-level) |
— |
Every variant runs at or above 95% utilization on at least one budget (HyperCook on mass/power/cost/volume simultaneously; EverSimmer on cost). LeanBroth is the only variant with comfortable margin (47.8-60.0%) across all four resource budgets.
All eight SR gates pass for all three variants — every variant is a compliant candidate baseline; the trade study exists to rank compliant designs, not to eliminate non-compliant ones.
| Gate | SR | HyperCook | LeanBroth | EverSimmer |
|---|---|---|---|---|
| Mass ≤ 15,000 kg | SR-GS-011 | PASS | PASS | PASS |
| Power ≤ 500 kW | SR-GS-012 | PASS | PASS | PASS |
| Cost ≤ 2,000 kCr | SR-GS-013 | PASS | PASS | PASS |
| Volume ≤ 400 m³ | SR-GS-014 | PASS | PASS | PASS |
| Throughput ≥ 200 bph | SR-GS-002 | PASS | PASS | PASS |
| Automation ≥ 0.8 | SR-GS-003 | PASS | PASS | PASS |
| Operators ≤ 5 | SR-GS-004 | PASS | PASS | PASS |
| Gravity ≥ 12 g | SR-GS-015/016 | PASS | PASS | PASS |
| All 8 gates | PASS | PASS | PASS |
Min-max normalized scores (see 05_trade_study_methodology.md §3) show a consistent pattern: EverSimmer leads on Automation, Availability, and N-1 Retention (it is the only variant with any single-fault throughput retention at all — the other two both retain 0% after their worst-case single-unit loss); LeanBroth leads on ResourceMargin and CostMargin by a wide margin; HyperCook leads only on ThroughputMargin, and even there its lead over EverSimmer is modest (+60% vs. +20% margin) given EverSimmer's parallel-cell topology also produces real throughput headroom.
| Scenario | HyperCook (A) | LeanBroth (B) | EverSimmer (C) | Winner |
|---|---|---|---|---|
| Balanced | 0.342 | 0.347 | 0.671 | EverSimmer |
| ThroughputFirst | 0.467 | 0.297 | 0.585 | EverSimmer |
| CostLean | 0.196 | 0.615 | 0.514 | LeanBroth |
| MissionAssurance | 0.242 | 0.312 | 0.812 | EverSimmer |
EverSimmer wins three of the four named scenarios, including ThroughputFirst — a scenario weighted 35% toward raw throughput margin, where HyperCook was expected to dominate. HyperCook's razor-thin margins on every other criterion (cost, resources, automation headroom, N-1 retention) drag its weighted score down even under throughput-weighted scoring, so its throughput edge alone is not enough to win. LeanBroth wins only under CostLean, the one scenario that weights cost and resource margin most heavily (0.55 combined) — consistent with it being purpose-built as the resource-budget-optimized variant.
Across 5,000 random Dirichlet weight draws (rng(42), see methodology §3), the win shares are:
| Variant | Win share |
|---|---|
| HyperCook (A) | 5.0% |
| LeanBroth (B) | 11.0% |
| EverSimmer (C) | 84.0% |
EverSimmer wins 84% of random weightings spanning the full space of plausible stakeholder priorities — not just the four hand-picked scenarios above — indicating the result is not an artifact of scenario selection.
HyperCook (A). Delivers the highest raw throughput (320 bph, +60% margin) but at the cost of running within 0.4-1.0% of the mass, power, cost, and volume caps simultaneously — the least margin of any variant on four of five budget dimensions at once. Its single-string cook/prep/transport topology also gives it 0% N-1 throughput retention, matching LeanBroth's worst case despite HyperCook being the "high-automation, high-throughput" design. It loses even the ThroughputFirst scenario to EverSimmer because near-zero margin everywhere else swamps its throughput lead once any other criterion is weighted at all.
LeanBroth (B). The clear resource-budget leader — comfortably under every mass/power/cost/volume cap (47.8-60.0% utilization) and lowest absolute cost (1,070 kCr). It meets, but does not comfortably clear, its other targets: throughput margin is only +5% (210 vs. 200 bph floor) and automation is exactly at the 0.8 floor with zero margin. It wins the CostLean scenario and takes a real (11%) share of the Monte Carlo draws, confirming it is the right choice specifically when budget headroom is prioritized over throughput or resilience.
EverSimmer (C). Wins 3 of 4 named scenarios and 84% of random weightings — the most robust performer across the plausible range of stakeholder priorities. Its triplicated-cell topology is the only one of the three with any graceful degradation at all (66.7% retention, 160 bph after losing one cell), and it leads on automation (0.956) and availability (0.9789) as well. Its principal weakness is cost: at 1,905 kCr it is 95.3% of the cost cap, second only to HyperCook, with just a 4.7% cost margin — the resilience and autonomy gains are not free.
- EverSimmer's cost margin is only 4.7%. At 1,905 of 2,000 kCr, EverSimmer has the least cost headroom of any compliant variant apart from HyperCook. Any further requirement growth, vendor cost increase, or triplicated-cell scope creep is a real risk of pushing it over the SR-GS-013 cost cap.
- EverSimmer's degraded-mode throughput (160 bph) is below the SR-GS-002 nominal floor (200 bph). The 66.7% N-1 retention figure satisfies the intent of SR-GS-026 (no uncontrolled production halt from a single fault) as a graceful-degradation contingency mode, but 160 bph is not a compliant steady-state operating point against SR-GS-002. Degraded operation following a cell loss must be treated and documented as a contingency mode, not an alternate compliant nominal state.
- LeanBroth passes automation (SR-GS-003) with exactly zero margin. At 0.800 average automation against a 0.8 floor, any single component substitution, added manual step, or measurement rounding could tip LeanBroth into non-compliance. This is the tightest gate margin observed across all three variants on any SR.
- HyperCook is one requirement-growth event away from multiple simultaneous violations. Its power (0.4% margin), cost (1.0% margin), and volume (0.7% margin) margins are all under 1%. A single component overrun in any of these dimensions — not unusual during detailed design — would likely cascade into cap violations across more than one budget at once, since HyperCook has essentially no slack to absorb it in any of the three.
Adopt EverSimmer (Variant C) as the baseline physical architecture.
EverSimmer wins 3 of 4 stakeholder weighting scenarios (Balanced 0.671, ThroughputFirst 0.585, MissionAssurance 0.812 — all highest of the three) and 84% of 5,000 Monte Carlo random weightings, the most robust result of any variant across the plausible range of stakeholder priorities. It is the only variant offering any graceful degradation under single-fault conditions (66.7% N-1 retention vs. 0% for both HyperCook and LeanBroth), leads on automation and availability, and clears all eight SR compliance gates.
This recommendation carries three follow-up actions to address the caveats above:
- Negotiate a cost reserve or descope items to widen EverSimmer's 4.7% cost margin before detailed design locks in triplicated-cell vendor costs, since it is currently the second-tightest cost margin of the three variants.
- Define a degraded-mode operations procedure for the 160 bph single-cell-loss contingency, documenting it explicitly as a below-nominal contingency state (not a compliant steady-state alternative to the 200 bph SR-GS-002 floor) with clear criteria for when to invoke it and how operations and customer commitments adjust while in it.
- Carry LeanBroth as a documented descope option. If budget priorities shift and mass/power/cost/volume margin becomes the dominant driver, LeanBroth's CostLean-scenario win (0.615) and 11% Monte Carlo win share make it the next-best alternative to re-evaluate, given it comfortably clears every resource budget where EverSimmer and HyperCook do not.