Container Home CO Detectors

A shipping container is a very good structural object and a very poor ventilation object. It was designed to keep seawater out for a month at a time, and it does that job by being sealed. When you cut a window into one and move in, you inherit a building envelope that has almost no accidental air exchange — and you inherit it in the smallest volume most people will ever live in.

Almost every piece of mainstream carbon monoxide advice assumes the opposite conditions: a house of roughly eight to sixteen thousand cubic feet, with enough leakage through the shell that a small combustion mistake gets diluted before it becomes a problem. A container home violates both assumptions at once. This article is about what actually changes, and it is written on the assumption that a compliant CO alarm layer is already installed — if you have not read what the UL 2034 standard actually tests, start there, because everything below is about the gap *above* that baseline.

The Volume Problem Comes First

A standard 20-foot ISO container has interior dimensions of roughly 19 feet 4 inches by 7 feet 8 inches by 7 feet 10 inches. That is about 1,170 cubic feet, or 33 cubic metres, of total air volume — before you subtract the insulation, the cabinetry, the bed platform and your own body. A high-cube unit buys you another foot of height and lands near 1,320 cubic feet.

Compare that with the house the guidance was written for. A modest 2,000-square-foot home with eight-foot ceilings holds about 16,000 cubic feet. A container home is somewhere between one twelfth and one fourteenth of that.

This matters because indoor CO concentration is, to a first approximation, a source emission rate divided by a dilution rate. Halve the air volume and the same appliance fault produces roughly twice the concentration. Drop the air change rate at the same time and the peak gets higher still. The U.S. Consumer Product Safety Commission's own modelling of furnace emissions makes this explicit: in their scenarios, moving from a 100-square-metre home at 0.35 air changes per hour to a 200-square-metre home at 0.7 air changes per hour cut projected indoor CO concentrations by about 75 percent. The two variables that produced that swing were volume and ventilation — and a container home sits at the far end of both.

None of this means a container home is intrinsically unsafe. It means the margin between "a small combustion fault" and "a meaningful indoor concentration" is much thinner than in a house, and that the assumptions baked into generic advice no longer hold.

Where CO Actually Comes From in a Container Build

Container homes skew off-grid, and off-grid heating and cooking skews combustion. That is the whole story in one sentence. Here is the detail:

SourceWhy it is common in container buildsWhat makes it different at 33 m³
Propane cooktop or rangeCheapest reliable off-grid cooking; no electric capacity neededTransient spikes from a single burner are diluted in a house; here they register
Propane tankless water heaterCompact, wall-mounted, no tank volumeDraws large combustion air flow; if the flue or intake is compromised the room is the dilution volume
Absorption (propane) refrigeratorRuns continuously, no compressor, silentA fault is not an event — it is a background level that never resets
Unvented propane or kerosene heaterCheap, needs no flue penetration through steelEmission goes straight into the living volume, with no roof cut required
Small wood stoveOff-grid heat, off-grid fuelFlue must penetrate a steel roof; wind-driven downdrafting is a real risk on an exposed site
Portable generatorStandard off-grid battery chargingUsually staged outside, but "outside" is often a lean-to or the lee side of a door
Vehicle or equipment started nearbyRural sites, no second vehicleIdling exhaust enters through the one open door or a window on the lee side

The pattern to notice is not the sources themselves — every one of these appears in conventional houses too. It is that in a container build, several of them are often running at once in a volume with no spare air. That is the scenario the false-alarm and background-level analysis describes as a raised baseline rather than a single event.

If you are weighing a propane system specifically, our RV propane detector guide covers the separate question of fuel-gas detection, which is a different problem from CO and needs its own sensor.

Propane Sinks. Carbon Monoxide Does Not.

This is the single most useful physical distinction to internalise, and it is the one most people get wrong.

Propane is heavier than air. In a sealed steel box, a leak pools at floor level and travels along the floor to the lowest point, which in a container is usually under the bed platform or behind the cabinetry at the far end from the doors. WorkSafeBC's guidance on repurposed containers is blunt about this: the units are designed to be watertight, they are well sealed with little or no ventilation, and that is ideal for shipping and potentially dangerous for other uses. Their concern is ignitable vapour accumulation, not CO — but the air-movement behaviour they describe is the same behaviour that governs where a fuel-gas sensor should sit.

Carbon monoxide does not behave that way. NFPA 72's annex material notes that CO has a density very close to air, which is why CO alarms are neither low-level nor high-level devices in the way a propane sensor is. CO mixes. Within a 33-cubic-metre volume, it mixes fast — a single-room air volume that small reaches a fairly uniform concentration in minutes rather than the tens of minutes a house might take.

The practical consequence is a pair of contradictory-looking conclusions that are both correct: put your fuel-gas detection low, because propane sinks; put your CO detection at breathing height in the space where people actually are, because CO is already everywhere. Our placement height guide goes through the reasoning in more detail for conventional rooms.

Spray Foam, Steel, and the End of Accidental Ventilation

Houses leak. Old houses leak a lot: through sheathing seams, around window frames, through attic penetrations, through the endless small gaps a wood-framed building accumulates. That leakage is not comfortable and it is not efficient, but it does provide a baseline dilution that CO guidance quietly relies on.

A container does not have that. The shell is welded corrugated steel. The corrugation is not decoration — it is what gives a thin steel plate the rigidity to be stacked nine high at sea, and it means the envelope is continuous in a way a framed wall can never be. Most conversions then add closed-cell spray foam directly to the inside of the steel, which is the right call for condensation control on a thermal bridge this aggressive, and which also happens to make the envelope airtight by accident rather than by design.

The result is a building with essentially zero infiltration. Every air change has to be deliberate: a window, a door, a vent, or a mechanical system that is switched on. That is genuinely good for energy use and genuinely demanding for combustion safety, because any appliance that needs combustion air now has to be given it explicitly rather than borrowing it from the shell.

This is also why the temperature and humidity monitoring angle matters more than it sounds. In a steel box with spray foam, condensation is a structural issue, not just a comfort one, and humidity is the proxy you can actually watch.

Negative Pressure Arrives in Seconds, Not Minutes

Exhaust fans cause reverse venting. This is not a container-specific observation — the standard CO alarm literature lists negative pressure differential from exhaust fans as one of the recognised causes of transient CO problems, alongside wind effects and simultaneously running appliances.

What is container-specific is the time constant. Depressurising a house enough to backdraft a water heater takes a sustained run of a powerful fan. Depressurising 33 cubic metres takes a small bath fan or a range hood on high for a fraction of that time, because there is so little air to move and nowhere for replacement air to come from.

The failure sequence looks like this: you run the range hood while cooking on a propane cooktop. The hood pulls air out faster than it can be replaced. The container goes slightly negative. The nearest combustion appliance — a tankless water heater on the same wall, or a wood stove — now has its flue working against a reverse pressure gradient, and its combustion products come into the room instead of going up the pipe. You created the fault by doing the correct thing.

The fix is combustion air and interlocked makeup air, and it is a design-stage fix rather than a detector fix. But it is worth knowing which of your normal actions are capable of producing it, because a CO event that correlates perfectly with "every time I cook" is almost always this, and it is diagnosable in a way that a one-off event is not. See furnace and boiler CO behaviour for the same mechanism at conventional scale.

Wood Stoves and Flues Through Steel

The small wood stove is the most popular off-grid heat source in container builds and the one with the most installation risk, for a reason that has nothing to do with the stove.

To vent it, you cut a hole in the roof of a steel box. That penetration has to maintain clearance to combustible material — which, in a spray-foam-lined container, is the foam itself and any interior cladding — and it has to do that while the steel expands and contracts through a much wider temperature range than a wood-framed roof. A steel roof on an exposed rural site in full sun reaches temperatures a shingle roof never approaches, and it cools just as fast at night. That movement works on every seal at the penetration.

Then there is wind. Container homes are frequently sited in open ground, which is exactly where wind-driven downdrafting is strongest. A flue that drafts correctly on a still day can reverse in a gust if the termination is poorly sited relative to the roof geometry.

None of this is an argument against wood stoves. It is an argument for knowing that a stove in a container has more ways to fail than a stove in a house, and that the failure produces CO in the smallest dilution volume available.

What UL 2034 Does and Does Not Do for You

This is the section that most container home owners have not been told, and it is the reason a display matters more here than in a house.

UL 2034 is a time-and-concentration standard. The alarm is required to sound within these windows when exposed to these concentrations:

CO concentrationRequired alarm response windowRequired resistance (must NOT alarm)
70 ppmBetween 60 and 240 minutesMust not alarm for 60 minutes
150 ppmBetween 10 and 50 minutesMust not alarm for 10 minutes
400 ppmBetween 4 and 15 minutesMust not alarm for 4 minutes
30 ppmOptional pre-alarm allowed after 1 hourMust not alarm for 30 days

Two things follow directly from that table, and both are by design rather than by defect.

First, the resistance requirements exist to stop nuisance alarms. A compliant unit is *required* to stay silent at 30 ppm for thirty days. That is correct behaviour for a house, where 30 ppm might be a nearby road or a normally-operating appliance, and where a nuisance alarm trains people to remove batteries.

Second, and this is the container problem: in a volume this small, the concentrations that sit below the mandatory alarm window are the ones you are most likely to live with. A slow continuous source — an absorption refrigerator with a marginal burner, a heater that is slightly out of adjustment, a flue that downdrafts only at night — can hold a space in the tens of ppm for hours without ever crossing a mandatory alarm threshold. The alarm is behaving exactly as designed. It is also telling you nothing.

This is not a criticism of the standard. It is a statement about what layer the standard operates on. Our safe ppm level chart lays out the same thresholds with the interpretation attached.

The 21-Foot Rule Stops Being the Constraint

NFPA 72's CO alarm placement language includes a requirement to install outside a 21-foot radius of fuel-burning appliances. Note what that actually is: an installation distance, not a coverage radius. A detector does not cover 21 feet in any meaningful sense; the rule exists to keep the device out of the immediate, turbulent, high-concentration zone right at an appliance where it would nuisance-alarm.

Now apply it to a container. The interior length of a 20-foot unit is 19 feet 4 inches. The *entire interior* is inside 21 feet of a cooktop mounted at one end. There is no compliant location, because there is nowhere far enough away.

The correct response is not to despair about code. It is to recognise that the rule was written for a building type where "away from the appliance" means something, and that in a single-volume 33-cubic-metre space the meaningful placement question is different: put the CO detection where the breathing happens, avoid the immediate splash zone directly above the cooktop, and accept that distance is no longer a variable you control. Our how many detectors you need guide covers the general reasoning; in a container the answer is usually "at least one in the sleeping volume, regardless of what the count rule would suggest for the square footage."

Where it differs from our barndominium guide is that a barndominium has the opposite geometry problem — a large building where distance and separate air volumes are the whole difficulty. A container has one air volume and no distance. Both break the default rule, in opposite directions.

Why the Number Matters More Than the Horn Here

Put the last three sections together and the conclusion is fairly direct.

In a house, a CO alarm is a binary device attached to a large, forgiving volume. It is silent for years, then it screams, and the scream is the whole product. That is a reasonable design because in a house the interesting events are rare and large.

In a container home the events you should care about are more often small and sustained: a refrigerator burner drifting, a heater slightly out of adjustment, a flue that only misbehaves in a north wind, a generator positioned thirty feet upwind of the only open window. Those events live in the band between background and the mandatory alarm window. They are exactly the events an alarm layer is required to ignore.

A number changes what you can do about them. Not because a number is more accurate than a certified alarm — it is not, and it is not trying to be — but because a trend tells you *which* of your appliances is the problem and *when* it is the problem. A reading that creeps from single digits to the twenties every evening when the water heater runs is diagnostic information. It tells you to look at that flue, at that moment, under those conditions. A horn that never sounds tells you nothing at all, which is precisely what a compliant alarm will do in that scenario.

This is the argument for a plug-in unit with a live reading as a second layer rather than a replacement: keep the alarm layer doing what it is certified to do, and add the observability that the alarm layer is explicitly designed not to provide. That framing is the whole point of a digital display detector, and it is worth reading that piece before deciding where the display goes.

Off-Grid Power Has a Specific Compatibility Trap

Container homes are disproportionately solar-and-battery, and there is a compatibility requirement here that almost nobody mentions until something fails.

Carbon monoxide alarm documentation is consistent on this point: alarms intended for AC power should be operated only with true or pure sine wave inverters. Operating them from a square wave or quasi-sine-wave inverter — which includes a large share of consumer UPS products and the cheaper pure-power inverters — can damage the alarm. A damaged alarm does not necessarily stop working in an obvious way, which is the part that should concern you.

This is worth checking at design time rather than at failure time, and it applies whether the unit is a standalone alarm or a plug-in monitor. If your inverter is not documented as pure sine wave, find out what it is before you depend on anything plugged into it. Our power outage guide covers the related question of what happens when the supply disappears entirely.

There is also a plain logistics point in favour of plug-in units in a container: the volume is small enough that a single outlet position usually serves the whole space, and a unit that also reports temperature and humidity earns its outlet twice in a steel box where condensation is a standing concern. A plug-in monitor with battery backup covers both the compatibility question and the outage case in one device.

A Practical Setup for a Single 20-Foot Unit

The following is a configuration that fits the physical reality described above. It is not a substitute for whatever your jurisdiction requires — container homes sit in an inconsistent regulatory space, and some areas treat them as accessory structures with no residential CO requirement at all.

  1. Keep a compliant CO alarm layer. This is the layer that is designed and tested to wake you. Nothing below replaces it.
  2. Place it in the sleeping volume, not in the kitchen end. In a single-volume space, "sleeping volume" means where your head is at night, away from the immediate cooktop splash zone.
  3. Add a fuel-gas sensor low. Propane sinks. A CO sensor at breathing height will not find a pooled fuel leak, because CO and propane are different problems with different densities.
  4. Add a read-out layer. Put it where you will pass it every day, at a height you can read without crouching. A display you never look at provides nothing — and in a build this small, one plug-in unit with a live CO reading usually covers the whole volume from a single outlet.
  5. Give combustion appliances explicit air. If you have a flued appliance and a mechanical extractor, understand what happens when both run. This is the highest-value thing on the list and the only one that is not a purchase.
  6. Check the flue penetration annually. Steel moves. Seals at a roof penetration are the most likely place for a slow degradation you cannot see from inside.
  7. Log your own baseline. Spend a week noticing what the reading does when you cook, when the water heater fires, when the stove is lit. That baseline is what makes an anomaly recognisable.

For the alarm-vs-monitor distinction in more depth, plug-in versus battery units is the comparison, and our combination detector buying guide covers what to look for when one device is doing both jobs.

FAQ

Do I need a CO detector in a container home if I have no gas appliances?

If nothing in the building burns fuel, the indoor source risk is close to zero. Most container builds have at least a cooktop, a water heater, a fridge or a stove, and most also have a generator somewhere nearby. Check the generator location specifically — generator CO guidance applies even when the unit is outdoors.

Is a container home worse than a tiny house for CO?

Volume-wise they are comparable, but the envelope is not. A stick-built tiny house leaks; a welded steel box with closed-cell foam effectively does not. The dilution you get for free in a tiny house has to be designed for in a container. Our tiny house guide is the nearest comparison.

Where exactly should the detector go if the whole container is within 21 feet?

In the sleeping volume, out of the immediate zone directly above the cooktop. The 21-foot language assumes a building with somewhere farther away to put it; a 20-foot container does not have one. See the placement height guide.

Will a standard CO alarm catch a slow leak from a propane fridge?

Not reliably at low concentrations, by design. A compliant unit must not alarm at 30 ppm for thirty days and must not alarm at 70 ppm for the first hour. A slow continuous source can sit in that band indefinitely. That is the specific gap a readable number covers.

Why does the reading spike when I run the range hood?

That is depressurisation reversing a flue. It is one of the recognised transient CO mechanisms listed in alarm documentation, and it happens much faster in a small sealed volume than in a house. Treat it as a combustion-air problem, not a detector problem.

Can I run a plug-in monitor from my solar inverter?

Only if the inverter is documented true or pure sine wave. Square wave and quasi-sine-wave outputs can damage alarms and monitors, and the damage is not always visible. Verify the inverter specification rather than assuming.

Does CO sink to the floor like propane?

No. CO has a density close to air and mixes through the volume. Propane is heavier than air and pools low. They need detection in different places, which is why one device at breathing height is not a complete answer for a propane-fuelled build.

How often should I test?

Test the alarm layer on the schedule in its documentation, and check the flue penetration at least annually. Beyond that, our testing guide has the procedure, and how long detectors last covers end of life.

What should I do if the alarm sounds?

Get out and get fresh air, then deal with the cause from outside. Our beeping alarm guide covers what different patterns mean, and CO poisoning symptoms has the symptom reference. This article is not medical guidance and nothing here replaces a professional assessment of your specific installation.

Is a three-pack overkill for a single container?

For one 20-foot unit, probably — but multi-packs exist because most builds are not one container. The common pattern is a 40-foot main volume plus a 20-foot addition used as a bedroom or utility space, and that is two air volumes with different sources, which is the case how many you need is written for.

Where a Second Layer Fits

The honest summary of everything above is that a container home does not need different CO equipment so much as it needs a different mental model. The alarm layer is the same. The volume is not, the envelope is not, and the events worth catching are concentrated in the band a compliant alarm is required to ignore.

What closes that gap is not a better alarm — it is observability. A unit that shows the live CO reading, tracks the peak while you were out, and also reports the temperature and humidity that drive condensation on a steel shell gives you something a horn structurally cannot: the ability to correlate a number with an action you took. For a build where the appliance, the flue and the sleeping position are often within six feet of each other, that correlation is the whole game.

Our plug-in unit is built as a supplemental monitor for exactly this purpose — CO, combustible gas, smoke, temperature and humidity on one screen, with battery backup for the off-grid case where power is the first thing to go. It is not a replacement for a certified alarm, and it does not describe itself as one. It is the layer that shows you the number before the number becomes an alarm.

If your build runs on propane at all, pair it with a proper fuel-gas detection setup, because propane pools low and CO does not — and no single sensor height handles both. And if you want the reasoning behind why an alarm can stay silent and still be working correctly, the UL 2034 explainer is the piece that makes the rest of this make sense.

Related Reading

Back to the blog title
0 comments
Post comment
Note: comments needs to be approved before publication

Cart

loading