A continuous emissions monitoring system, commonly abbreviated to CEMS, is an integrated arrangement that measures source emissions repeatedly or continuously over extended operation. It normally includes analysers, probes or optical paths, sample-handling components, calibration facilities, a data acquisition system and supporting measurements used to interpret the pollutant signals. Its value lies not in a single reading, but in a time-resolved record that shows how emissions respond to changes in production, combustion, fuel, control equipment and plant condition.
A complete system extends beyond the analyser cabinet. The measurement chain begins where the gas is observed or extracted and continues through any probe, filter, heated line, cooler, pump, converter or dilution stage. The analyser converts a physical or chemical response into a concentration signal, while the data system stores that signal together with timestamps, operating status and quality-control information.
Supporting instruments may measure oxygen, moisture, temperature, pressure or gas flow where those variables are needed to place the pollutant result on the required reporting basis. The separate flue-gas-monitoring page explains oxygen correction, moisture and reference conditions; this page focuses on how continuous systems obtain, verify and manage the underlying measurements.
System design should begin with the source and the intended determinands. Gas composition, temperature, pressure, dust loading, moisture, corrosive constituents and concentration range all influence the chosen equipment. A system suitable for a clean combustion exhaust may not remain reliable on a wet, particle-laden or chemically reactive process stream.
Non-dispersive infrared analysers measure the absorption of infrared radiation at wavelengths associated with particular gases. They are widely applied to gases with suitable infrared absorption characteristics, although water vapour, carbon dioxide and other matrix components may create interference if the optical and conditioning design does not address them. Multi-channel instruments can measure more than one component, but each channel still requires its own performance checks.
Ultraviolet absorption and ultraviolet fluorescence are used for gases that interact strongly with ultraviolet radiation. Chemiluminescence analysers measure light produced by a controlled chemical reaction and are commonly used for nitrogen oxide measurement. Where different nitrogen species must be represented together, the system may include a converter whose efficiency becomes part of the quality-assurance programme.
Paramagnetic and zirconia instruments are established approaches for oxygen measurement, while electrochemical cells may be used in some system designs where their range, response and replacement characteristics are suitable. Flame ionisation detection is used for total gaseous organic carbon or hydrocarbon response because carbon-containing compounds produce ions in a hydrogen flame. The reported signal is method-dependent and should not be interpreted as a compound-specific analysis.
Fourier-transform infrared spectroscopy can measure several gases from a single spectrum and is particularly useful where a multi-component measurement is required. Optical transmissometers estimate opacity or light attenuation across a stack, while triboelectric and related particle-monitoring devices respond to interactions between particles and a probe. These particle instruments can provide valuable trend or control information, but their signal does not automatically equal a reference-method particulate concentration without an appropriate relationship and quality framework.
An extractive system removes a sample from the source and transports it to an analyser. This arrangement allows instruments to be installed in an accessible, temperature-controlled location and can support several analysers from a common sampling system. It also creates additional surfaces and components where the sample may cool, condense, react, adsorb or become diluted.
Extractive systems are often described as hot-wet, cold-dry or dilution configurations. A hot-wet arrangement keeps the sample above the relevant condensation temperature and measures the moist gas without deliberate water removal. A cold-dry system cools the sample and removes condensate before analysis, while a dilution system mixes a controlled amount of clean gas with the extracted sample so that the diluted mixture can be transported and measured under less demanding conditions.
An in-situ analyser measures directly within or across the stack without transporting the complete sample to a remote cabinet. Probe-type instruments observe gas near a defined insertion point, while cross-stack optical systems send radiation across a measurement path. In-situ arrangements reduce sample-transfer losses, but the optical path or probe remains exposed to heat, dust, vibration, deposits and alignment changes.
Neither configuration is universally superior. The decision depends on pollutant behaviour, source conditions, access, maintenance capability, response-time needs and the ability to perform meaningful calibration checks. A hybrid arrangement may use in-situ measurement for one determinand and extractive analysers for others.
Sample conditioning should make the gas suitable for analysis without removing or changing the determinand. A probe filter may prevent particles from entering the sampling line, but the filter material and temperature must be compatible with the gases being measured. Heated probes and lines are used where cooling could cause water or another component to condense before reaching the analyser.
Cold-dry systems normally include a controlled cooling or permeation stage to reduce moisture. Condensate management is important because blocked drains, flooded coolers or unstable cooling can alter the sample and damage downstream equipment. The system should identify whether a result is produced on a dry or moist basis and should retain the information needed for any subsequent conversion.
Pumps, flow controllers and pressure regulators maintain consistent transport through the system. Leaks can draw ambient air into a line under suction, diluting the sample and changing oxygen-dependent results. Restrictions can slow response, increase residence time and create differences between the actual source condition and the value displayed at the analyser.
Reactive or soluble gases require particular attention to material selection and residence time. Stainless steel, glass, fluoropolymer and other materials have different adsorption and chemical properties. A conditioning train should therefore be assessed as a complete measurement path rather than assembled from individually convenient components.
Calibration establishes the relationship between analyser response and a known input. Zero gas provides a reference corresponding to the absence, or defined minimal presence, of the measured component. Span gas introduces a certified concentration within the instrument's working range. Calibration materials, delivery systems and pressure conditions should be suitable for the analyser and traceable through documented certificates and identifiers.
A routine zero and span check tests whether the analyser remains close to its established response. Drift is the change between the expected and observed response over time. A failed drift check may indicate contamination, optical deterioration, detector instability, converter failure, calibration-gas problems or a fault in the sample path. Adjustment should not erase the evidence that the drift occurred.
Additional performance checks may examine linearity, response time, interference, converter efficiency, leak integrity and agreement with an independent measurement. The relevant checks depend on the analyser principle and system design. Automated checks can improve consistency, but automatic valves and calibration sequences also require verification because a successful software command does not prove that the correct gas reached the measurement cell.
Calibration and maintenance activities should be recorded with date, time, person or automated routine, gas-cylinder identity, expected value, observed response, adjustment and outcome. The record should make it possible to determine which emission data may have been affected before or after a failed check.
The data acquisition and handling system receives analyser signals and combines them with timestamps, source status, calibration events and fault codes. It may calculate averages, apply conversion factors and produce reports, but the unprocessed or minimally processed signal should remain retrievable. An unexplained final value is difficult to audit if the original measurement and calculation history have been overwritten.
Data validation separates usable emission measurements from periods affected by maintenance, calibration, instrument fault, loss of sample flow or invalid source status. The rules should be defined in advance and applied consistently. Validity flags should not be changed merely because a result appears unusual; unusual values may represent a genuine process event and should be investigated against plant records.
Data availability describes the extent to which the system produced valid data during the period in which measurement was expected. A high volume of stored values does not necessarily mean high availability if the values were generated during analyser faults or without a representative sample. Availability reporting should therefore distinguish valid measurement, quality-control activity, planned maintenance, source downtime and unplanned system loss.
Time synchronisation is also important. If analyser, process-control and reporting systems use different clocks, an emission peak may be matched to the wrong operational event. Clock changes, communication interruptions and manual data entry should leave an audit trail rather than silently shifting or replacing records.
Preventive maintenance should address the entire measurement chain. Typical tasks include inspecting probes, cleaning optical surfaces, replacing filters, checking heated lines, servicing pumps, clearing drains, examining gas connections and reviewing analyser diagnostics. Maintenance frequency should reflect source severity and observed performance rather than relying only on a generic calendar.
Technical review should compare emission trends with process operation and system status. Sudden changes may arise from a real process condition, control-equipment behaviour, a blocked filter, a leak, calibration-gas depletion or analyser interference. The reviewer should use diagnostic and process evidence before classifying the event.
A CEMS report should identify the source, determinands, analyser principles, measurement configuration, reporting basis, data period, validation rules, calibration and maintenance activity, system outages, data availability and any substituted or estimated values. Graphs can help show trends, but they should preserve units, time scale and status information. Reports should distinguish measured data from calculated, corrected, substituted or manually entered data.
Long-term records support comparison across operating periods and can reveal gradual deterioration that a short campaign may miss. They do not remove the need for periodic independent stack testing where that work has a separate purpose. The relationship between a continuous system and a reference test should be defined rather than assuming that either form of measurement automatically replaces the other.
CEMS produces a time-resolved record across extended plant operation, while periodic stack testing measures selected conditions during defined test runs. The two approaches can answer different questions and may use different measurement principles.
No. Analyser suitability depends on the physical and chemical behaviour of each determinand, the source matrix, expected range and potential interferences. Multi-component instruments can measure several gases, but they still require determinand-specific validation.
A heated line can prevent water or another component from condensing before the sample reaches the analyser. Condensation may remove soluble gases, alter concentration, block the line or damage equipment.
Drift is a change in analyser response relative to a known zero or span input over time. It can indicate deterioration, contamination, instability or a problem elsewhere in the measurement path and should be investigated and documented.
No. Stored values may coincide with calibration, maintenance, sample-flow loss, analyser fault or an invalid source condition. A validation system should assign status information and retain the reason for excluding or qualifying data.
It should identify the affected periods, explain the cause where known, distinguish planned and unplanned loss, state how availability was determined and clearly label any substituted or estimated values.