Carbon monoxide detectors in HUD housing: where they go and how many you need

Carbon monoxide alarms are now one of the most consequential items in a HUD assisted unit. Federal law requires HUD assisted housing to meet recognized carbon monoxide alarm standards, and under NSPIRE a missing or inoperable alarm is treated at the highest severity level, which means a correction window measured in hours.

The rules themselves are short. Applying them across a portfolio of different building types is where the work is.

Where alarms have to be

The recognized standards place alarms in two kinds of locations.

Outside each separate sleeping area, in the immediate vicinity of the bedrooms. A unit with bedrooms grouped off a single hallway needs an alarm in that hallway. A unit with bedrooms in two separate parts of the home needs an alarm outside each grouping.

On every level of a multi level dwelling unit, including habitable basement levels. A two story townhome does not satisfy the requirement with a single alarm on the bedroom level.

Manufacturer instructions govern the exact mounting. Alarms have specified distances from fuel burning appliances, from cooking appliances, from vents and from corners, and installing inside those exclusion zones causes nuisance alarms that residents then disable. A disabled alarm and a missing alarm score the same.

How many you need

Count sleeping area groupings first, then count levels, then take the higher number. A single level unit with all bedrooms off one hall needs one. A single level unit with a primary suite at one end and two bedrooms at the other needs two. A three level unit needs one per level regardless of where the bedrooms are.

Combination smoke and carbon monoxide alarms satisfy both requirements when they are listed for both purposes, which is why most portfolios standardize on one device model. Standardizing also simplifies the part of this that actually consumes budget, which is replacement.

Which units are covered

The requirement attaches to dwelling units in covered HUD assisted properties. In practice, owners should assume every unit is in scope rather than trying to exempt units, because the exemption analysis turns on fuel fired appliances, attached garages and adjacency to sources, and getting it wrong on one unit is a life threatening finding on inspection day.

Properties that are all electric with no attached garages and no fuel fired equipment anywhere on site have a narrower obligation, but the analysis still needs to consider shared mechanical spaces and any generator or boiler serving the building.

Alarms expire. A device installed at the same time as the rest of the portfolio will reach end of life at the same time as the rest of the portfolio.

The part portfolios forget

Carbon monoxide alarms have a manufacturer end of life, commonly in the range of five to ten years depending on the model, after which the sensor is no longer reliable and many devices chirp and shut down. An expired device is not a compliant device.

Build a device inventory with model, install date and end of life date, by unit. Replace on a schedule rather than on failure, because failure arrives in batches and inspection notice does not wait for a purchase order.

Document annual testing. Test dates, unit numbers, and the name of the person who tested. When an inspector finds an inoperable alarm, the difference between a program failure and an isolated device failure is whether you can show the testing record.

A ninety minute action list

Pull your unit mix and count required devices by floor plan. Compare that count to what purchasing says is installed. The gap is your immediate exposure.

Pick a random sample of ten occupied units, verify placement against the rules above, press the test button, and note end of life dates. Ten units will tell you whether you have a device problem, a placement problem, or a records problem.

Then work the rest of the life safety category the same way, which is the sequence we use in NSPIRE inspection readiness engagements and describe in our overview of what changed from REAC to NSPIRE.

This is the cheapest severe finding to prevent in the entire standard. It is also the one most frequently cited, because nobody schedules it until somebody fails it.