QAL2 and the Calibration & Variability Test: certifying your CEMS after install

Operating a compliant CEMS — Part 2 of 5. A practical series for industry, drawn from the DOE CEMS Guidelines (Version 8, 2025), which we helped develop.

In Part 1 we mapped the four-stage quality-assurance chain. QAL1 proved the instrument design is sound before it ever reached your stack. But a certified analyser bolted onto your chimney, breathing your flue gas, still has to prove it reads true. That proof is QAL2 — and until it passes, your data officially counts for nothing.

What QAL2 actually certifies

The QAL2 Calibration and Variability Test (QAL2-CVT) is the test that accepts and certifies your CEMS installation and operation at the plant. It does two jobs at once: it calibrates your analyser against a trusted reference, and it checks that the calibrated system is precise enough to be relied on.

Here’s the part operators most often underestimate: QAL2 must be completed before your CEMS data is connected to the DOE System for CEMS (the iRemote platform). Any data transmitted before that is, in the words of the guidelines, “unverified and invalid.” QAL2 is the gate between installed and officially reporting.

First, the functional test

Before any of the real measurement happens, the system has to clear a functional test — a series of checks (alignment, zero and span, leak, interferences, drift) confirming the CEMS is fit to be tested. It must be carried out no more than one month before the parallel measurements begin, and validated before the test proceeds. If it fails and no corrective action is taken, the QAL2 simply cannot go ahead. It’s the gate we flagged in Part 1 — the thing that stops you wasting a test campaign on a system that isn’t ready.

The parallel measurement: CEMS against the SRM

This is the heart of QAL2. A DOE-registered, independent CEMS tester runs the Standard Reference Method (SRM) — the trusted manual measurement method (what the SRM actually is, and why your CEMS depends on it) — alongside your CEMS, sampling the same flue gas at the same time. Pair after pair, the SRM gives the “true” value to compare your analyser against.

The guidelines are specific about how this is done:

• At least 15 valid parallel measurements, spread uniformly over three days (not necessarily consecutive, but within four weeks), with the plant operating normally.
• Each sample at least 30 minutes, and the start of each at least one hour apart.
• At least one measurement at or near zero (defined as ≤ 5% of the ELV).
• Testers are advised to collect 18–19 sets so enough valid pairs remain after any outliers are removed.

One rule matters more than it looks: reference gases alone are not permitted to build the calibration. The test has to use the real stack-gas matrix — only the SRM, on your actual emissions, captures the moisture, temperature and interferences your analyser really sees. The tester also picks times when emissions are likely to be highest and most varied (without artificially forcing them), so the calibration covers your genuine operating range.

The calibration function

From those paired readings, the tester derives the calibration function — the equation that converts your analyser’s raw signal into a true, traceable concentration. From this point on, your CEMS no longer reports raw counts; it reports calibrated, defensible values.

A correlation of R² ≥ 0.9 is the guideline’s indicator of a good fit. But where emissions are low or tightly clustered, a lower R² can be perfectly legitimate — which is exactly why R² is guidance, and the variability test has the final word.

The valid calibration range

A calibration is only trustworthy across the span of concentrations it was actually built on — the valid calibration range (VCR). The guidelines set its top at the highest calibrated parallel reading plus a margin (10% for gas, or 100% for particulate, of the ELV — or 20% of the ELV, whichever is greater).

Operate inside the VCR and your data stands on solid ground. The operator must check weekly that readings stay within it; readings that persistently sit outside the VCR (for example, more than 5% of values over five consecutive weeks) can trigger a fresh QAL2 — though if an exceedance was caused by a plant failure, fixing the plant is enough and a full recalibration isn’t needed.

The variability test — the decider

Calibration tells you the CEMS reads the right value. The variability test asks whether it reads consistently enough. It compares the scatter of CEMS-versus-SRM differences against an allowed uncertainty (the 95% confidence interval in the regulation), using stack measurements only — never reference materials.

This is the definitive test. A CEMS can have a tidy-looking calibration line and still fail variability; the test, not the eye, decides. Worth knowing too: sloppy SRM work can make a perfectly good CEMS fail — another reason the competence of your independent tester is not a place to economise.

Passing — and going live

Once the system passes, the tester compiles the Calibration and Variability Test report (to the Appendix 2 format) and submits it to DOE for verification. Only after that verification can your CEMS data connect to the DOE System for CEMS as valid, compliant data.

And then the job shifts from proving it once to keeping it true every day — which is QAL3, the subject of the next part.

Facing a QAL2 and want it to pass first time? Talk to us — we prepare and run CEMS for facilities across Malaysia against the same DOE CEMS Guidelines we helped write.

This article is general guidance, not legal advice. For obligations specific to your facility, refer to the current Environmental Quality (Clean Air) Regulations 2014, the EQA 1974, and the DOE CEMS Guidelines, or speak with us directly.