DOE · Intensity–mass chart · Claim gate · Arm compare · Tracker export · History
Matches ANC Fuse oil catalog. Notes + vapor pressure drive Raoult blend safety and baseline decay.
Matches production blend: FO + DPG + IPM + Benzyl Benzoate + odor eliminator. Affects retention and flash risk.
Models forced-convection boundary layer thinning ($h_m$) when hanging in front of active vehicle A/C vents.
Top / Middle / Base note distribution. Auto-loads from the selected fragrance; adjust to model custom balances. Drives chart curves, component half-lives, composite \(k\), and scent character.
Fits scale weigh-ins to linear regression on ln(M(t)/M₀) to derive the empirical base rate constant. Keep at least 2 points.
| Day (t) | Scale Weight (g) | Oil Mass (M_t) | Action |
|---|
Score 0–5 on the same days you weigh (match customer tracker). Noticeable = score ≥ 3. Builds pilot calibration without external panelists.
Checklist before Save / Freeze. Claim level unlocks copy strength.
Left axis: pilot intensity (0–5). Right axis: oil mass remaining (% of M₀ from weigh-ins). Horizontal line at intensity 3 = noticeable threshold. When intensity falls below 3 while mass remains, residual oil is no longer “product.”
| Day | Total (g) | Top (g) | Mid (g) | Base (g) | Scent Character |
|---|
Claim gate: Heuristic → qualitative only. Pilot intensity → screening note. Full panel (≥5) → Dlabel = 0.8 × min(panel median, hot-car median).
Downloads a lab handoff file for you to update the customer tracker baselines. Does not upload to live users.
Compares the latest saved Control and Hot-car runs for the oil selected below (or current oil). Claim inputs can use min(control, hot-car).
Save multiple runs, then click Best-in-Class to rank by lifespan, R², cost/day and FAI.
Human validation of intensity, character & perceived longevity. Supports blinded / randomized codes.
Factorial designs, RSM planners, and next-run suggestions from your history.
Adjust hypothetical settings and predict noticeable horizon / composite k using the physics model + ML when trained.
Client-side random-forest ensemble trained on your LocalStorage run history. No cloud required.
Plan a central composite or Box–Behnken layout around current settings for continuous factors.