Corrected, dry and referenced: when a dust reading becomes a compliance number

Drawn from the DOE CEMS Guidelines (Version 8, 2025), which we helped develop, the international standards they implement — EN 14181 and ISO 9096 — and the Environmental Quality (Clean Air) Regulations 2014. A plain walk-through of why the number on a dust monitor is not yet the number the law compares against — and what closes the gap.

A transmissometer dust CEMS on a biomass boiler — the kind sitting on almost every palm oil mill stack in Malaysia — will, after calibration, put a tidy figure on the screen: so many milligrams per cubic metre. It is tempting to read that figure straight across to the limit — 150 mg/m³ — and call it compliant or not. But the two numbers are not on the same footing. The limit in the Clean Air Regulations is not “150 mg/m³” full stop; it is 150 mg/m³ at standard temperature and pressure, on a dry basis, referenced to 12% carbon dioxide. The reading coming off a bare dust monitor is at none of those conditions. It is measured in hot, wet flue gas, at whatever pressure and whatever CO₂ the boiler happens to be producing that minute.

Getting from one to the other is what “correction” and “normalisation” mean. It is not a formality, and — this is the part that matters for a great many mills — it needs measurements a dust monitor does not make on its own.

And here the field reality is stark. In Malaysian palm oil mills today, the transmissometer dust monitor is installed on its own. The moisture and CO₂ measurements that normalisation depends on are simply not there, and in most cases no data-acquisition correction stage is present at all. So the number on the screen is, and stays, a reading at actual stack conditions — while the limit it will be judged against is written on a different basis entirely. This article is about that gap: what it is, why it is easy to miss, and what closes it.

What the limit is actually written against

Every emission limit has to state the conditions it applies at, or it can be gamed. Open a damper, pull in more excess air, and the same mass of dust leaving the stack is diluted into a larger gas volume — the concentration falls without a single milligram less being emitted. So the regulation fixes the conditions: report the dust as if the gas were dry, at 273 K and 101.3 kPa, and diluted to a fixed reference level. For fuel-burning equipment not covered by the First Schedule — which is where an ordinary biomass boiler, a palm oil mill’s included, sits — the Second Schedule of CAR 2014 is explicit: “The CO₂ reference content is 12%,” and Total Particulate Matter is limited to 150 mg/m³.

That single sentence settles a question we are asked constantly on site: should the correction gas be CO₂ or O₂? For a biomass boiler under the Second Schedule, the schedule has already answered — it is CO₂, at 12%. Oxygen only enters for the large installations in the Third Schedule (Heat and Power boilers above 10 MWe, referenced to 6% O₂ for solid fuel). The two are not interchangeable house preferences; the regulation assigns a different reference gas to each category, and for the mill boiler it is carbon dioxide.

From a light measurement to a compliance number

A transmissometer dust monitor never measures milligrams — it measures light, and its control unit turns that light into a concentration using the calibration coefficients cc0, cc1 and cc2 established during QAL2. That is the whole job of the control unit, and it is done superbly. But notice what basis the number lands on. The calibration is built against gravimetric sampling expressed at the conditions the monitor itself sees — the actual, wet, hot stack gas — so the mg/m³ the control unit outputs is dust at actual stack conditions. Honest, traceable, correctly calibrated — and not yet on the ELV’s basis.

Three more steps stand between that reading and a compliance number, and each one needs a measurement the dust monitor does not make:

The sensors that do the correcting

The correction is not something the dust monitor’s control unit does, and never was. Under the Guidelines, converting the reading to STP, drying it, and referencing it to 12% CO₂ is a data-acquisition function — the job of the data-acquisition and handling system (DAHS), which would take the dust reading and combine it with live gas measurements. On a dust-only mill that stage has nothing to work with — and, in practice, is usually not even present — because the sensors that feed it were never installed.

Four measurements are involved. Temperature and pressure bring the gas to STP; a modern in-situ dust monitor can usually supply these itself. The two that a dust-only mill is genuinely missing are moisture (to reach a dry basis) and CO₂ (to reach the 12% reference). Those are the additions that actually move the number.

There is a tempting shortcut worth naming so nobody takes it: fitting the control unit’s cc0/cc1/cc2 against standardised sampling instead, so the monitor appears to output a “12% CO₂” number with no peripherals at all. That does not measure anything — it bakes the average CO₂ and moisture of the calibration day into the coefficients, freezes them there, and lets them drift as the boiler’s excess air and fuel wetness change through the season. Worse, it hides the assumption inside the sensor’s output, where there is no separate raw record to check it against. The honest calibration and the correct engineering are the same choice here: let the monitor output real dust at measured conditions, and do the normalisation downstream where it can be seen.

Why the annual test passes anyway — and why that is the trap

Here is where mill operators often expect trouble that does not come, and miss the trouble that does. If the installation cannot normalise, surely the QAL2 calibration and the Annual Surveillance Test (AST) — which compare the CEMS against fresh reference sampling — must fail?

They do not, and the reason is instructive — and worth being precise about, because none of this is a DOE peculiarity. The QAL2 and AST procedure is EN 14181, the international standard the DOE Guidelines implement, and the reference sampling it is measured against is the gravimetric method of ISO 9096 (aligned in Malaysia as MS 1596). During any QAL2 or AST campaign the reference tester brings a full set of portable instruments and measures the stack’s temperature, pressure, moisture and CO₂ for every run — EN 14181 requires it. Those readings describe the stack gas, which the CEMS is looking at simultaneously, so the same figures are used to standardise both sides. And the calibration function itself, by EN 14181’s method, is fitted with the reference values converted to the monitor’s actual measuring conditions — not standardised — so the coefficients are legitimate regardless. The variability and acceptance tests then run on standardised values on both sides.

Now the subtle part. Because the same correction factor is applied to the CEMS value and the reference value — they are the same gas at the same moment — the normalisation scales both together and preserves their agreement. A monitor that tracks the reference well at stack conditions tracks it just as well after both are standardised. Normalisation cannot make a well-calibrated instrument fail. So the dust-only CEMS passes QAL2 and passes AST.

And that is the trap, because passing is not the same as reporting correctly. The tester plugs the sensor gap for one day and leaves. The other 364 days, with no CO₂ and no moisture on the stack, the DAHS — where there is one — has nothing to normalise with, and the value transmitted to DOE’s CEMS system every minute is dust at actual conditions — compared, at DOE’s end, against a limit written at 12% CO₂. The annual test certifies a referenced measurement the installation only assembles on test day.

What this means for a mill

First, check which regime the stack is even in. The continuous-CEMS obligation is triggered when the boiler’s dust load reaches 2.5 kg/h — a figure read off the stack-test certificate as concentration times flow, not a desk estimate. If Total Particulate Matter for that specific premise is periodic-only under its licence, there is no continuous calibration to maintain and no permanent peripherals are needed: each periodic isokinetic test measures and corrects its own result on the day. The four-sensor question only bites where continuous monitoring applies.

Where it does apply, the honest position is straightforward, and it is two claims kept firmly apart. The calibration can be stated with full confidence: the monitor’s cc0/cc1/cc2 are properly derived and traceable to corrected, normalised reference sampling. What cannot yet be claimed is that the daily reported value is on the 12% CO₂ basis — that waits on live CO₂ and moisture measurement feeding the DAHS. A referenced instrument on paper and a referenced number every minute are different things, and only the sensors close the gap between them.

So the case for adding a QAL1-certified CO₂ analyser and a moisture measurement is not about passing the annual test — a dust-only mill already does. It is about making the everyday number mean what the annual test certifies, and what the limit assumes. That is the difference between a monitor that looks compliant once a year and a CEMS that reports compliantly every minute. It is, in the end, the whole point of continuous monitoring.

If you are running a dust-only transmissometer on a biomass boiler and are not sure which side of this line you are on, talk to us — establishing the correct correction chain, and the sensors it needs, is core to what we do.

This article explains how the DOE CEMS Guidelines and the Clean Air Regulations 2014 treat correction and normalisation of dust data. It is general guidance, not a substitute for your stack’s licence conditions or a DOE-registered CEMS tester’s assessment of your installation.