Skip to content
BAYHEAT

Lab report BH-001 · the 4× problem

One garage. Seven answers. 13,200–51,300 BTU/h.

Rules of thumb can’t see your garage. Our open model can.

BayHeat IR · Modeled · 160×120
24×24×9 ft · attached · out 3.3°F · in 55°F
71°F / 0°F
As-is
9.3 kW
2 circuits × 30 A · 10 AWG
Heat escapesYour garage: DHolds heat
Size my garage — 60 s

No sign-up. No email. Just your ZIP.

FIG. 1 Published rules of thumb for the same garage, and our modeled band.
ASHRAE · climate dataEIA · gov’t fuel pricesNEC · electrical codeIFGC · gas codeNFPA · fire codeUL · safety listing

02 / Where your heat goes

The ceiling loses more than the walls. The door loses more than the ceiling.

FIG. 2 Our example garage: 24×24×9, attached, R-13 walls, uninsulated ceiling, one steel door.
Ceiling
34%
Air leaks
24%
Big door
23%
Slab edge
9%
Walls
8%
Windows
3%
House wall
-3%
Service door
2%
FIG. 3 Where your heat goes, sorted largest first.

03 / Can I run it?

Three real questions. Three real answers.

Big Buddy propane heater, attached garage, running overnight?

NO-GO

2 conditions to check

5 kW electric heater, a spare 240 V/30 A circuit, no gasoline stored?

GO · PER MANUAL

2 conditions to check

Diesel air heater, exhaust routed outdoors, CO alarm in the house?

ONLY IF

5 conditions to check

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

Check my heater →

04 / The Garage Heat Index

The US median garage costs $746 a season to heat on electric resistance.

Same standard 24×24×9 garage, held at 50°F, priced at each state’s own current electricity rate. Seals and a door kit cut the median about 47%. Savings vary. Find out why in the seller's fact sheet on R-values. Higher R-values mean greater insulating power.

Cheapest to heat

  • FL — Miami · $0.00/season
  • HI — Honolulu · $0.00/season
  • CA — Los Angeles · $2.00/season

Costliest to heat

  • AK — Anchorage · $3,652/season
  • ME — Portland · $2,453/season
  • VT — Burlington · $2,252/season
See all 51 states →

07 / How we work

Seven things we hold ourselves to.

  1. 01We never claim a test we didn't run. Measured means a published log, an instrument and a receipt.
  2. 02We buy every unit we test, at retail, and post the receipt with the address redacted. A pick without an M mark was chosen from manufacturer specs and our model — we have not tested it.
  3. 03No money changes a result, a ranking, a verdict or a model constant.
  4. 04No display ads — not on the planner, the reports, safety pages or money pages.
  5. 05We publish "Don't buy" verdicts, negative results and our model's misses.
  6. 06We correct within 72 hours, or 24 for safety. Every change goes into the Lab notebook.
  7. 07We show how every page is made: model version, sources, AI assistance, and who checked it.
Read the full standards page →
Size my garage — 60 s