Continental Furnaces Industrial Insights (Morning Edition): Emissions Monitoring That Holds Up 2026: CEMS Selection, Sampling-Site Rules, EN 14181 Calibration Discipline and Data Availability for Steel Rolling Mills, Heat Treatment Furnaces and Melting Operations

13 min read

A permit number is not a monitoring system. The stack, its analyser and its data trail are what make emissions performance observable, and where continuous reporting is required, the record is examined minute by minute. A furnace can be operating steadily and still produce a weak compliance record if the sampling plane is unrepresentative, the certified range is unsuitable, or the operator has not maintained QAL3 control charts.

This is a guide to monitoring and measurement quality, not emissions chemistry or combustion control. It is distinct from our 4 November annealing-atmosphere and emissions article, 14 November combustion-air and fan-systems article, 12 November OT and cybersecurity article, and 1 November waste-heat and flue-gas-condensation article. Standards such as EN 14181 and EN 15259 are published reference frameworks, not universal legal requirements. Confirm the monitoring route, standards and reporting duties required by your own permit and regulator.

1. Three ways to watch a stack, and where each one gives way

The right monitoring route depends on the permit, the pollutant, operating pattern and required data record.

  • Periodic manual sampling: A sample is extracted and analysed on site or in a laboratory. A sample may take 30 minutes to several hours. Useful for periodic verification and for measurements that do not require continuous reporting. It provides snapshots, not a continuous operating record.
  • Periodic instrumental monitoring: A transportable automated system is brought to the stack for a monitoring campaign. It can provide instrumental measurements for the test period, but is not a permanently installed monitor.
  • CEMS, in situ: The analyser measures in the stack or duct. This avoids a conventional extractive sample line, but alignment, optical-path conditions and access for checks matter.
  • CEMS, extractive: A probe draws sample gas to a permanent analyser at or near the stack. It provides continuous measurement, but every part of the sample path can affect the result.

No route is automatically superior. Select one that meets the permit’s monitoring objective and can be maintained and validated under real furnace conditions.

2. The sampling plane decides the answer before the analyser does

A high-performance analyser cannot correct a sample that is not representative. Environment Agency guidance for EN 15259 recommends locating the plane as far as possible from disturbances such as bends and fans, in a duct with constant shape and cross-section. The recommended position is at least 5 hydraulic diameters upstream and 2 downstream of disturbances, or at least 5 hydraulic diameters from the stack top.

For CEMS locations and grid measurements, the mandatory flow criteria are:

  • Flow angle below 15 degrees to the duct axis, with no local negative flow.
  • Minimum velocity corresponding to 5 Pa differential pressure, about 3 m/s.
  • Highest-to-lowest velocity ratio below 3:1.

These rules are consequential in furnace layouts. Short runs between a recuperator outlet and stack may leave little straight duct; an ID fan can introduce swirl; expansion joints and changes in section can disturb flow. Assess the stack, duct condition and port design together. Continental Furnaces’ refractories information is relevant when reviewing refractory-lined duct sections and the practical details of expansion joints and port installation.

A single gas sampling point is acceptable only after a homogeneity test demonstrates that concentrations are uniform, or a representative point is established. The test need not be repeated unless fuel type, stack structure or abatement systems change. Install ports before commissioning wherever practicable: retrofitting can require a plant shutdown. Guidance recommends ports of at least 125 mm internal diameter, with a 75 mm socket acceptable on stacks under 0.7 m diameter, and a permanent platform designed for the monitoring task.

Sampling port and permanent platform on a furnace plant stack

3. Selecting the analyser for furnace gas, not for a catalogue

The sample-handling method is part of the measurement instrument. Selection should match the certified system to expected gas conditions and documented site constraints.

  • Hot-wet extractive: Keeps the sample path hot rather than removing moisture through a chiller. The complete system still needs testing for leaks, response time and performance at the probe. Condensed moisture is especially important for soluble gases: EPA material cites published findings of 3% to 15% SO₂ loss in a chiller at 100 to 1,000 ppm, with a projection of about 30% loss at 10 ppm and 20% moisture.
  • Cool-dry extractive: Sample conditioning can introduce bias. A partly blinded probe filter can scrub SO₂ and make readings systematically low, even when plugging is not obvious. Leaks in negative-pressure lines dilute the sample. Adsorption on tubing walls is strongest for NO₂, then SO₂, then NO; O₂, CO and CO₂ do not commonly adsorb.
  • Dilution systems: Check dilution-air quality and regulated pressure. Readings can respond to stack absolute pressure, about 1% per 3.45 in. H₂O, and temperature, about 1% per 50°F drop, with a non-linear temperature effect. Multi-component calibration blends can introduce molecular-weight effects of up to 7%; a subsonic orifice changes the dilution ratio.
  • In-situ systems: Point monitors can foul at the probe filter, with response time lengthening first. Audit gas flow that is too high can pressurise a cavity and bias calibration; too little may not flush it. Path monitors provide a line average, which is not automatically an area-weighted average in a circular stack. Internal calibration cells can leak or contain decomposed gas, producing a silent calibration bias.

CEMS analyser shelter interior with sample lines and secured calibration gas cylinders

4. The certified range is an accuracy decision

A certified range is not just a display setting. MCERTS performance specifications are expressed as a percentage of that range, so a lower certified range generally gives better absolute performance. For example, a cross-sensitivity limit of ±4% means ±3 mg/m³ on a 0 to 75 mg/m³ certified range, but ±8 mg/m³ on a 0 to 200 mg/m³ range.

The guidance’s range-selection rule is that the lowest certified range should be no more than:

  • 1.5 times the daily average ELV for incineration processes.
  • 2.5 times the daily average ELV for large combustion plant and other process types.

Separately, set the on-site measurement range from the ELV and expected peaks, for example twice the short-term ELV, not from the lower values observed during QAL2. Capping peak readings is not acceptable except in specific exceptional circumstances agreed in advance with the regulator.

5. Monitoring routes compared

Route What it gives you Typical furnace fit Principal failure mode QA burden
Periodic manual sampling Test-period samples Periodic checks or permit-defined campaigns Snapshot misses operating variation Repeat test planning and reference method
Periodic instrumental (transportable) Automated measurements during a campaign Testing or temporary monitoring Campaign not representative Campaign checks and suitable reference method
Extractive CEMS Continuous sampled gas record Permanent monitoring where sample path can be maintained Sample loss, leakage or conditioning bias QAL1, QAL2, QAL3 and AST
In-situ CEMS Continuous measurement at the stack Permanent monitoring where location and optics suit Fouling, alignment or calibration-cell bias QAL1, QAL2, QAL3 and AST
PEMS Model-based emission estimate Only where validated and accepted for the installation Model or input sensor no longer representative Installation-specific validation and ongoing checks

6. The QA ladder, and the dates that catch people out

Under EN 14181, the assurance sequence for permanently installed compliance CEMS is:

  • QAL1: The measuring system is certified for the pollutant and suitable range, for example through MCERTS certification.
  • QAL2: Installation and calibration are checked against standard reference methods, alongside functional tests. At least 15 valid paired data points are required, including one at or near zero. Near zero means no more than 5% of the daily average ELV.
  • Functional tests: Complete them no more than one calendar month before parallel reference tests. Linearity uses five reference concentrations including zero, at about 20%, 40%, 60% and 80% of a range at least as large as the short-term ELV. Use at least 3 readings per concentration and 6 at zero.
  • NOx converter: Test efficiency at least annually. It must be at least 95%, with testing per EN 14792; gas-phase titration is recommended.
  • Test gases: Use traceable gases with uncertainty better than ±2% at 95% confidence. A useful starting point for span gas is 80% to 100% of the short-term ELV.
  • QAL3: The operator tracks ongoing drift and precision using control charts and warning and action limits.
  • AST: The annual surveillance test is a mini-QAL2, typically using 5 parallel reference measurements.

QAL2 is required at installation and at least every 5 years, plus or minus 2 months. AST is due every 12 months, plus or minus 2 months. Implement QAL2 results within 6 weeks of the report and no later than 6 months after the test date.

7. Data availability is the number that gets audited

QAL3 is about recognising genuine drift, not correcting every fluctuation. Set warning and action limits as multiples of the standard deviation, or SAMS. Do not adjust zero and span unless readings repeatedly breach action limits. The CEMS must display negative values so that readings around zero remain visible. Shewhart or CUSUM charts are common; CUSUM requires weekly zero and span checks.

A cylinder change can look like analyser drift. Take at least 5 span readings with the old cylinder, then at least 5 with the new cylinder, and reset the baseline. Keep QAL2 and AST raw data for at least 5 years. Auto-corrections are appropriate only when the system is certified for that approach and the corrections are recorded.

EN 14181 does not cover the data acquisition and handling system, or DAHS. Validate the DAHS and its data chain separately, including signal transfer, timestamps, scaling and records.

For Indian real-time reporting under Maharashtra Pollution Control Board guidance, transmission is at 1-minute intervals, with each average built from 10-second polls. Data must carry quality codes, including calibration mode, calibration drift and corrected datum. Upload lag must be under 15 minutes; otherwise post the data as delayed with the delayed-data quality code. In practice, an analyser fault and a network outage can produce the same visible gap to a regulator. Build outage detection and escalation into the operating procedure.

8. Normalising the number: oxygen, moisture and reference conditions

Comparisons are valid only when results share the same reference conditions. Common oxygen reference levels are 3% for gas or liquid fuels, 6% for solid fuels, 11% for most incineration processes and 15% for gas turbines.

The oxygen correction factor is:

Fo = (21 − reference O₂%) / (21 − measured O₂%)

For dry-gas conversion:

dry = wet × 100 / (100 − water%)

For ppm-to-mg/m³ conversion, use the pollutant’s molecular weight and a molar volume of 22.40 litres at 273 K and 101.3 kPa. Calculate mass emission rate as concentration multiplied by volumetric flow, with both values at the same reference conditions.

High oxygen can create apparent ELV breaches and apparent QAL2 or AST failures, particularly during batch-process start-up or shutdown. The guidance notes that O₂ above 18% can be considered abnormal. Review the operating phase and permit conditions before interpreting the result.

9. When a CEMS is not the defensible choice

PEMS may be an option only when the model and its inputs are proven for the specific installation. An EPA example protocol for an industrial furnace model used feed type, firing rate, fuel-gas density, furnace air-preheat temperature, excess oxygen, stack temperature, inlet-air humidity and inlet-air temperature.

That example set relative-accuracy targets of NOx within 20% of the average test-method value or 10% of the applicable standard, whichever is greater, and CO within 10% of the average test-method value or 5 ppm, whichever is greater. It used 9 test data sets across low, normal and high firing rates. Ongoing checks included quarterly electronic data accuracy within ±5% of expected output, sensor validation against a model of each sensor built from the other inputs, semiannual relative-accuracy tests, or annual tests if the most recent result was 7.5% or better, and valid data for at least 95% of unit operating hours.

The Environment Agency notes that field evidence supports PEMS for NO and CO, especially for smaller gas turbines below 50 MW thermal input. Evidence for other installations and pollutants is limited, so each installation needs proof and validation. PD CEN 17198 addresses PEMS software certification and performance assurance. For dust monitors where gravimetric calibration cannot be established, EN 17389 covers quality assurance for automated dust arrestment monitors, including filter dust and filter leakage monitors certified to EN 15859.

Close-up of extractive CEMS sampling probe and sintered filter assembly

10. The 2026 complication: a fuel change is a calibration event

A QAL2 must be repeated after significant repairs or upgrades affecting calibration, a change of fuel, a significant process change, or operation above the valid calibration range. A significant change to plant operation or fuel requires QAL2 within 6 months.

A fuel change is significant when it changes the emissions profile and requires a permit variation, or when switching between gaseous, liquid and solid fuel types, or between a single fuel and a mixture, with the alternative fuel used for more than 10% of the time in a year. Hydrogen blending and biofuel trials belong in the project trigger register: assess the fuel category, expected emissions profile and permit implications before the trial begins. Treat oxygen-enriched operation as a process-change review item where it may affect the calibration basis. This is capital-project planning, not a discussion of combustion chemistry.

11. Monitoring considerations across furnace applications

A steel rolling mill should plan stack monitoring around the duct arrangement and process phases; heat treatment furnaces need a measurement plan that remains valid across their operating pattern. An aluminum melting furnace and a melting furnace for steel require analyser and sample-path selection based on their permitted pollutants and stack conditions. An industrial furnace manufacturer should treat ports, access and data interfaces as design considerations, not late additions. Properly integrated industrial furnace systems align the monitoring arrangement with commissioning and lifecycle service. In the wire and cable industry, the same principle applies: define the measurement objective and reporting route at project outset. A hot dip galvanizing plant, metal recycling furnace and broader thermal processing equipment project each need a monitoring plan matched to the permit, not a generic analyser specification.

12. Keep the monitoring chain available, not just the analyser

A continuous record depends on consumables and support equipment throughout the chain. Review on-site stock and replacement plans for:

  • Probe filters and sinter elements.
  • Heated sample line and umbilicals.
  • Sample pumps and diaphragms.
  • Chiller and peristaltic condensate pumps.
  • NOx converter and zirconia O₂ cells.
  • Blowback solenoids.
  • Calibration gas regulators, cylinders and zero-air filters.
  • Port gaskets, caps and blind flanges.
  • DAHS computer and network hardware.

Continental Furnaces supports lifecycle availability through furnace spare parts and accessories. Match replacement items to the certified system and its approved configuration.

13. A five-phase roadmap

  1. Phase 1: obligation and permit matrix. List pollutants, limits, reporting intervals, methods and required regulator submissions.
  2. Phase 2: sampling plane and homogeneity assessment. Review duct geometry, flow disturbance, access, ports and representative-point evidence.
  3. Phase 3: analyser technique and certified range selection. Select in-situ, extractive or other defensible route, then match certification and measurement ranges to the ELV and expected peaks.
  4. Phase 4: QAL2, QAL3 baselines and control-chart limits. Complete functional checks, reference tests, baseline records and operator procedures.
  5. Phase 5: annual surveillance cycle, fuel-change trigger register and PEMS re-verification. Schedule ASTs, record process and fuel changes, and revalidate any model-based monitoring when inputs or operating conditions change.

14. Review the monitoring arrangement alongside the furnace

A dependable emissions record is engineered across the full system: furnace operation, stack and duct, sampling plane, analyser, DAHS and operator routines. Continental Furnaces brings 35+ years of expertise, ISO-certified quality, customized solutions and prompt service to support enduring customer partnerships and minimize downtime.

Invite our team to review your furnace and its monitoring arrangement together. Request a quote and consultation as a practical step toward sustained competitive advantage.

15. References

These standards and guidance documents are published reference frameworks, not universal legal requirements. Confirm the obligations in your own permit and with your regulator.

  1. Environment Agency, M20: Monitoring stack emissions, quality assurance of continuous monitoring, updated 14 September 2026.
  2. Environment Agency, Monitoring stack emissions: measurement locations, EN 15259 guidance.
  3. Environment Agency, Monitoring stack emissions: technical guidance for selecting a monitoring approach.
  4. US EPA, An Operator’s Guide to Eliminating Bias in CEM Systems, Chapter 3: Extractive Systems and Chapter 4: In-situ Systems.
  5. US EPA, Alternative Monitoring Protocol: Predictive Emission Monitoring System to Determine NOx and CO Emissions from an Industrial Furnace.
  6. Maharashtra Pollution Control Board, FAQ for Real Time Online Monitoring of the Effluent & Emission Monitoring System.

Ready to Optimize Your Thermal Processing?

Contact our experts for a free consultation.

Translate »
+91 98113 04306