Skip to content
G-005 · REV 1 · CHECKED 2026-10-02 · MODEL v1.0.2 ▾

Drafted with AI assistance. Numbers computed by model v1.0.2 or sourced. Human review: pending (BayHeat editorial desk). Licensed review: not applicable.

Technical review: no page-specific licensed approval recorded.

M measured · C computed · S spec · R reference · E estimate

The Lab

The Lab is where a hypothesis gets a published prediction before anyone measures anything. Every test below is pre-registered: the model's number goes up first, so a later physical result can be checked against it instead of quietly matched to it.

No physical test has run yet. Nothing on this page carries an M (measured) evidence mark. Everything below is R (a published rule, quoted) or C (computed by our model) — a prediction, not a result.

Test board

ProtocolHypothesisModel's predictionStatus
P-002A door kit plus weatherstrip, sealed together, cut heat loss enough to matter on their own.Door kit: (). Weatherstrip: (). Predicted independently — not a sum, since sealing both at once changes each other's draft path.PRE-REGISTERED
P-003A small diesel air heater isn't cheaper to run than electric resistance at typical prices, and it puts combustion byproducts into the garage.On this garage's design load: diesel about , electric resistance about , at this week's prices. CO output isn't modeled — that's what the test measures.PRE-REGISTERED
P-004A measured warm-up curve tracks the model's predicted curve within its stated error budget.For the sealed version of our standard example (door kit, weatherstrip and ceiling insulation applied), the model sizes a class and predicts to reach 55°F from a January cold start.PRE-REGISTERED
P-005An unvented propane heater (Big Buddy class) raises indoor humidity measurably compared to electric heat, since burning propane releases water vapor as a byproduct.No number predicted — our model doesn't compute humidity. This test sets the baseline instead of checking one.PRE-REGISTERED

Nothing running right now. Every row above is a prediction, waiting on an instrument.

Published so far

BH-001, "The 4× problem" is a computed report, not a physical test — it compares published sizing rules to our own model's spread across four real envelopes of the same garage. Every model release, price refresh and correction is dated in the Lab notebook.

Instruments on hand

None yet. This list grows the day a purchase receipt is posted, with its address redacted — not before. A phase-1 kit (thermal camera, a clamp meter, CO/CO₂ monitors) is planned but not yet bought.

Evidence marks

Every number on this site carries a mark: M measured (a published log exists), C computed by our model, S from a manufacturer spec, R a published reference or code citation, E a labeled estimate. Nothing on this page is marked M.

Safety, before any physical test runs

A combustion test (diesel, propane) runs only in a detached garage, with CO alarms placed in any adjoining building, and is logged remotely with the tester outside. Any detectable carbon monoxide ends the test immediately. No torpedo heater runs indoors, ever, on this site or off it. A protocol is reviewed before it's run, not after.

General information. Your electrician, gas fitter, local code and the heater's manual govern.

Report a problem with anything on this page — a wrong number, a broken link, a claim we should walk back — and it goes in the notebook within 72 hours, or 24 for a safety issue.

Related

Size my garage — 60 s