Flue gas analyser readings: how to take and read them

The analyser tells the truth about combustion only if the measurement itself is taken properly: at the sampling point the manufacturer specifies, on the right basis, and after the readings have had time to settle. A figure read too early, or at the wrong point, describes the measurement rather than the appliance.

What readings does a flue gas analyser give you?

Four figures matter on a typical domestic job: carbon monoxide (CO), measured in ppm; carbon dioxide (CO₂), measured as a percentage; the combustion ratio, which is CO divided by CO₂; and oxygen, with flue temperature alongside. Most analysers display the ratio directly, and it can also be worked out from the two readings: with CO in ppm and CO₂ as a percentage, the ratio is CO ÷ (10000 × CO₂), which is exactly what our free calculator does. Each figure answers a different question. CO and the ratio speak to how completely the appliance is burning, CO₂ and oxygen to how much air is in the mix, and flue temperature to what the appliance is doing with the heat it makes.

What do good flue gas readings look like?

Settled, inside the manufacturer's published figures for that model, and with a combustion ratio at or below 0.0040. There is no universal band, and a table that offers you one is selling a shortcut. What does exist is a two-layer answer: the generic key action levels Gas Safe Register's Technical Bulletin 143 gives where a manufacturer sets none, and the specific figures each maker publishes for each appliance. The table below shows both, using two current UK boilers as examples.

Read that third column as an example, not a standard. Every figure in it is the maker's natural gas value, and both manuals give different ones for LPG. TB 143 is explicit that the boiler manufacturer's guidance must be followed where it is available, and that makers may set their own, tighter action levels; both appliances above do. Flue gas temperature is the one reading with no commissioning band at all, and Worcester quotes the Greenstar 4000's separately for 80/60 °C and 40/30 °C operation, so what to expect there depends on the flow and return temperatures the appliance is running at. The figures for the boiler in front of you live in its installation instructions, and that book beats anything carried over from the last job.

What do the readings mean when they go wrong?

Read the figures together rather than one at a time, because each pattern points somewhere different. The table below sticks to the causes and actions TB 143 and the two manufacturers' instructions actually name.

What does a full reading set look like?

Here is a commissioning visit on a Greenstar 4000 combi, walked through against Worcester's published figures. The readings are invented for the example; the figures they are judged against are the manufacturer's.

Change one number and the visit changes. Had the minimum-rate CO₂ come back at 9.1%, it would be outside Worcester's 8.2 to 9.0% band and only 0.1 below the maximum-rate reading, failing both halves of the same requirement, and the boiler would not be signed off on it. A comfortable ratio does not rescue that, because the ratio is a screening figure and the maker's CO₂ window is a separate condition to satisfy.

What should you check before the probe goes in?

What is a dry flue gas reading?

Burning gas produces water vapour as well as carbon dioxide, so the gas in a working flue is wet: part of the sample, by volume, is steam. An electronic analyser does not read it that way. The sample is conditioned on its way to the sensors, with the water vapour condensed out and removed, so the figures on the screen describe the dry products of combustion. That is what “dry flue gas” means, and it is the basis manufacturers use for their commissioning data, which is what makes the comparison valid: a dry reading judged against dry reference figures is like against like. It also means a reading is not something you can sanity-check by eye against the plume at a flue terminal, because the visible cloud is mostly that condensed water, doing something the sensors never saw.

How do you sample a chimney or an open flue?

By the procedure in that appliance's instructions, at the sampling point the manufacturer provides. Where an appliance has no flue gas sampling point and the correct procedure is not specified in its manual, the manufacturers' own commissioning flowchart says to consult them for guidance rather than improvise a position, and on older open-flued kit that is often the honest answer.

The standard behind analyser use here is BS 7967:2015, which covers flueless (type A), open-flued (type B) and room-sealed (type C) appliances, along with any appliance for which the manufacturer has provided a purpose-designed sampling point or specific sampling instructions. It assumes an operative with the relevant ACS competence and equipment conforming to BS EN 50379-3 or BS 7927, and TB 143 points to it for further guidance on using an analyser. TB 143 itself does not stretch this far, and says so on its first page: its subject is commissioning a condensing boiler incorporating air/gas ratio control valve technology, and it explicitly does not apply to service and maintenance work involving that valve.

Probe position matters more in an open flue than anywhere else, because the flue is doing its job by drawing room air. Where a manufacturer gives a test point the instruction is specific: adjust the cone so it seals the point and puts the end of the probe in the centre of the flue gas stream. A probe sitting off-centre, or a point left unsealed, samples a blend of combustion products and room air. That blend lowers CO and CO₂ together, which is what makes a raw CO figure from an open flue such a poor witness, and why the ratio between them is the reading to lean on: dilution moves both of its ingredients at once, so the ratio holds. Oxygen is the tell. Well above what the appliance's data expects and the sample is diluted, whatever the CO figure says.

One check you have on a room-sealed appliance is missing here. The flue integrity check the manufacturers describe, oxygen of 20.6% or more with CO₂ below 0.2%, is taken at an air inlet test point, and an open flue has none. The flowchart covers that case directly: where no air inlet test point is provided, the analyser check is not possible and flue integrity is verified visually instead, before the combustion readings are taken. The spillage test sits alongside all of this rather than inside it. It belongs to the chimney side of the job, under BS 5440-1 and the appliance manufacturer's instructions, and it asks a different question: are the products of combustion actually leaving the room? The analyser asks how well the appliance is burning. Neither answer stands in for the other.

How should the measurements be recorded?

At the appliance, while the analyser is still connected, with the readings alongside the commissioning data they were judged against. On a new boiler the destination is the Benchmark commissioning checklist, carrying CO and combustion ratio at maximum and minimum rate: HHIC boiler manufacturers have agreed that completing it is a condition of warranty, and because commissioning documentation is a Building Regulations requirement, Gas Safe Register will defect an installation where those figures are not correctly recorded. Copying figures into the van later, or off a curling thermal printout at home, is where transposed digits creep into combustion ratios, and that is not a small class of mistake. Manifold's FlueScan reads CO, CO₂ and the ratio straight off the analyser screen with the phone camera for you to confirm, so the numbers on the record are the numbers the analyser showed, and this year's readings sit next to last year's when the appliance's story needs telling.