ECLSS Reliability Growth Simulator (2D)

PDCA Improvement Program

Improvement rate β 0.35

Initial failure scale λ₀ 0.060

Number of test phases 5

Time acceleration 40 hr/s

Fleet Status

Operating hours T

Current MTBF

— hr

Failures logged

PDCA cycles run

Test phase

1 / 5

Instantaneous λ(T)

— /hr

Corrective-Action Log

Program idle — waiting for first failure…

How it works

Real spacecraft life-support hardware (ISS ECLSS, Orion, future Mars habitats) is tracked with the Crow-AMSAA reliability-growth model, the standard math behind a Plan-Do-Check-Act improvement program run across a series of test phases:

N(T) = λ₀ · T^β            cumulative expected failures
λ(T) = λ₀ · β · T^(β-1)     instantaneous failure intensity
MTBF(T) = 1 / λ(T)          mean time between failures

Plan/Do — the system operates; failures occur as a non-homogeneous Poisson process with intensity λ(T).
Check — every failure is logged; a rack module flashes red on the grid.
Act — a corrective action (redesign, procedure fix, part swap) is applied, the module flashes green, and one more point is added to the growth curve.
Test phases — the program budget is split into the chosen number of equal-length phases (dashed lines on the growth chart); each phase boundary is logged as a milestone, matching how real TAAF (Test-Analyze-And-Fix) programs report progress phase by phase even though the underlying failure process is continuous.
β (growth rate) — how effective the corrective actions are. β close to 0 ≈ no learning (constant failure rate); β near 1 ≈ each fix sharply reduces future failures. Real aerospace programs typically land β ≈ 0.3–0.5.
λ₀ — how failure-prone the very first design iteration is, before any improvement cycles have run.

Because β < 1, MTBF grows as T^(1-β) — improvement continues but with diminishing returns per cycle, exactly the shape real reliability-growth test programs plot on log-log paper.

ECLSS rack module Failure detected Corrective action applied

MTBF growth Phase bound.