Overview
Starship Build Engineering System
A dependency-free, runnable build-engineering decision system for **Starship Ship vehicle assembly, built to demonstrate the core skills of a Lead Starship Engineer (Ship Vehicle Assembly)**: lean high-rate manufacturing, build scheduling, work-instruction structure, weld/dimensional acceptance, tolerance analysis, and design-for-manufacture simplification.
It models the role's exact scope — **Integrated Payload Volume (IPV), PEZ dispenser, Raceway, Tank Cleaning, and Megabay 2 Stacking** — and turns a production target into the engineering artifacts the lead owns.
Honest scope (read first): the vehicle model insrc/vehicle.mjsis
notional, self-authored engineering data — not SpaceX proprietary data or
measured production figures. No physical hardware was built. Seeproof/LIMITATIONS.md. What *is* verified is that the
engineering math is correct and reproducible (node verify.mjs, 18/18
hand-checkable checks).
What it computes
| Capability | What it answers |
|---|---|
Takt + line balancing (src/lean.mjs) | Given a rate target, what is the takt? How few workstations can hold the work content without violating precedence? What is the line efficiency, bottleneck, and throughput? |
Critical path (src/schedule.mjs) | What is the vehicle build lead time, and which activities set it (zero-slack critical path)? Where is there float? |
Process capability (src/quality.mjs) | Is a weld/feature capable (Cp, Cpk, expected PPM, sigma level) against print limits? |
Tolerance stack-up (src/tolerance.mjs) | Does an assembly fit? Worst-case vs statistical (RSS) variation of the resultant gap. |
DFM/DFA (src/dfm.mjs) | How many parts are theoretically necessary? What part-count and assembly-time reduction did a redesign achieve? |
Quick start
node run.mjs # full build plan at 8 vehicles/day, one 8h shift
node run.mjs --rate 6 --shifts 2
node verify.mjs # 18 independently hand-checkable checks
Open public/tool.html in any browser for the live Build Line Planner (drag the rate/shift sliders and watch the station layout, critical path, weld capability, fit, and DFM update).
Headline result (notional model)
Ranked-Positional-Weight balancing of the 16-operation build at 60-min takt uses 11 stations vs 16 for a naive one-operation-per-station line (31.25% fewer) and lifts line efficiency from 52.1% to 75.8% (+45.5%). Critical-path lead time is 56 h along A → B → E → F → G. All numbers regenerate from node verify.mjs.
Layout
src/ lean.mjs schedule.mjs quality.mjs tolerance.mjs dfm.mjs vehicle.mjs buildplan.mjs
run.mjs CLI build-plan demo
verify.mjs verification suite (writes verification-report.json/.md)
public/tool.html live Build Line Planner (no build step, no deps)
proof/ IRS_AUDITOR proof package (generated + curated)
Proof
This outcome carries a full IRS_AUDITOR proof package under proof/. See proof/EXECUTIVE_EVIDENCE.md for claims-vs-evidence, proof/REPRODUCE.md for exact commands, and proof/PROOF_DECISION.json for the gate's decision.