Measuring an Unknown Emission

01 The specification that does not exist

A recurring pattern in scrubber enquiries is that the requester has detailed knowledge of the process and virtually none of the waste gas. The production route, batch sizes, reagents, and shift pattern are all documented; the composition of the stream leaving the reactor is not. It would be wrong to interpret this as negligence. It is a direct consequence of how emissions are regulated.

Environmental law governs the outlet. Under Directive 2010/75/EU and its national transpositions, an operator is required to demonstrate that the concentration leaving the stack remains below a permitted value, and monitoring obligations are framed accordingly. Nothing in a permit specifies what arrives at the inlet of an abatement device, because nothing needs to: for compliance purposes, the inlet is irrelevant. For design purposes, it is the only quantity that matters.

That asymmetry is the problem in its entirety. A limit value is a ceiling on the outcome; a scrubber is dimensioned from the load. Between the two lies a body of information that, in the absence of measurement, is routinely substituted by a literature value, a figure drawn from a nominally comparable plant, or a supplier's assumption — each of which is a numerical conjecture. The function of an emission survey is to replace that conjecture with a specification.

This article concerns the stage prior to the existence of a scrubber. Two adjacent questions are treated elsewhere: demonstrating at commissioning that an installed scrubber achieves its guaranteed removal efficiency is the subject of simultaneous inlet and outlet measurement, and maintaining demonstrable compliance over the operational life of a plant is the subject of emission control for wet gas scrubbers. The three constitute a single sequence: characterise, verify, control.

02 Six quantities a design cannot do without

A wet scrubber is not selected from a catalogue; it is calculated and designed. Every principal dimension — column diameter, packed height, number of transfer units, absorbent chemistry, liquid-to-gas ratio, and material of construction — derives from a small set of inlet quantities. A survey that fails to deliver all six leaves at least one dimension resting on an assumption.

• Speciation. Which compounds are present, not merely the aggregate concentration. Solubility, reactivity, and the required absorbent chemistry all follow from molecular identity. A highly soluble acid gas, a non-condensable, and a reactive sulphur compound each require a distinct absorption column; no aggregate figure distinguishes between them.
• Concentration and its temporal distribution. A single value is insufficient. What is required is a profile: the baseline, the peak, the duration of the peak, and its recurrence frequency.
• Volumetric flow, referred to stated temperature and pressure. Flow determines column diameter and gas-side velocity; without it, no concentration can be converted into a mass load.
• Temperature and moisture content. These determine whether a quench stage is necessary, at what point the stream reaches saturation, and how much water condenses or evaporates within the column — shifts that alter the concentration presented to the absorption stage.
• Particulate load and size distribution. Mass loading sets the duty; size distribution determines whether a wet scrubber is the appropriate technology. This is treated separately in section 05.
• Trace constituents governing materials selection and fouling. Halides, sulphur species, condensable organics, and dissolved solids rarely dominate the mass balance, yet they govern alloy or polymer selection, blowdown rate, and cleaning interval.

03 The measurement plan follows the process, not the instrument

The sequence in which a survey is designed is frequently inverted. The instrument is selected first and the measurement plan is constructed around it, when the correct sequence runs from the measurement objective to the sampling strategy and only then to the analytical technique. EN 15259 formalises this requirement: it stipulates that the measurement objective be defined prior to execution, that the measurement cross-section be demonstrably suitable, and that the number and position of sampling points be determined by the homogeneity of the stream rather than by convenience of access. A sampling point located immediately downstream of a bend samples a stratified flow field; no instrument corrects for that condition after the fact.

Campaign duration is governed by the same logic. In a continuous process operating at steady state, a comparatively short campaign may be representative. In a batch process it cannot. Charging, reaction, stripping, cleaning, and idle periods each produce a chemically distinct gas, and the quantity governing the design is not the shift mean but the concentration prevailing during the phase of highest emission. A scrubber sized on the shift mean is undersized for the duration of every peak; one sized on the peak concentration alone is oversized for the remainder of the cycle. The design case is a decision about which of those two errors is acceptable, and it can only be made once the full concentration profile is known.

It follows that the process state must be recorded concurrently with the measurement. A concentration trace unaccompanied by a log of plant operating conditions is uninterpretable: a peak cannot be attributed to a cause, a baseline cannot be explained, and the profile cannot be extrapolated to any production rate other than the one that prevailed during the campaign. The process record is an integral component of the measurement, not ancillary context.

Measurement uncertainty warrants equally explicit treatment. Every measured value carries an uncertainty budget arising from sampling, calibration, cross-interference, and process variability; for automated monitoring systems subject to EN 14181, the permissible expanded uncertainty is itself a prescribed parameter. A survey report that presents values without an associated uncertainty estimate invites the designer to treat them as exact — which they are not, and which no measurement ever is.

04 What each principle resolves, and what it cannot

No measurement principle is universal. Each responds to a specific physical property and is consequently insensitive to any species that lacks that property. Selecting an analytical technique is therefore an implicit statement about what the analyst expects to find — an uncomfortable position when the composition of the stream is precisely what remains unknown. The limitations described below are structural rather than instrumental: they are not a function of instrument quality but of the underlying detection physics.

• Fourier-transform infrared spectroscopy (FTIR) resolves multiple compounds simultaneously from a single spectrum, which makes it the natural primary instrument for an uncharacterised gas stream. It responds exclusively to infrared-active molecules: homonuclear diatomics — N₂, O₂, H₂, and Cl₂ — undergo no change in dipole moment upon vibration and are therefore invisible to it, regardless of concentration. Water vapour and carbon dioxide absorb strongly across broad spectral regions and may obscure the weaker absorption bands of trace species. Because water-soluble compounds such as HCl, NH₃, and HF are lost to condensate on cool surfaces, the entire sampling train must be maintained above the acid and water dew points; the heated sample line is not an accessory but an integral component of the method.
• Non-dispersive infrared spectroscopy (NDIR) is robust, cost-effective, and well suited to quantifying a single component known in advance that possesses a spectrally distinct absorption band. It does not speciate and exhibits cross-sensitivity to water vapour and carbon dioxide. Applied to an uncharacterised stream, it answers a question that has not yet been formulated.
• Flame ionisation detection (FID) measures a sum parameter: total organic carbon, as defined in EN 12619. The response is approximately proportional to the number of carbon atoms; oxygenated compounds yield a systematically depressed response relative to their true mass concentration, so the reading represents a carbon equivalent rather than a direct mass measurement. The technique identifies no individual compound. Incorporation of a catalytic converter permits separation of the methane fraction from the non-methane fraction — a useful distinction, but not a speciation.
• Photoionisation detection (PID) is a screening instrument. It detects only those compounds whose ionisation potential falls below the lamp energy level, typically 10.6 eV — so methane (12.6 eV) and the lighter alkanes are not detected. Elevated humidity suppresses the response. PID is valuable for rapid source location and unreliable as a quantitative basis for design.
• Gas chromatography (GC) separates and identifies the individual compounds that sum parameters aggregate and constitutes the reference technique where molecular identity is required — the BTEX group being the most common application. Cycle times of the order of minutes characterise steady-state composition adequately but resolve fast transients poorly.
• Electrochemical cells are selective, compact, and inexpensive. They are also subject to zero drift, cross-interference from non-target species, and a finite operational service life. They are appropriate as a supplementary measurement channel for a single known component, not as the primary basis for stream characterisation.

The methodological consequence is as follows: a sum parameter bounds the total contaminant load but identifies no individual species, while a speciation technique names the compounds present but may be insensitive to whatever lies outside its detection window. A survey built on either approach alone addresses only half the measurement objective. The project record substantiates this conclusion without requiring further argument.

Of the 46 documented scrubber projects, 39 record the analysers employed. FTIR was the most widely applied technique, appearing in 23 of those 39 projects, followed by FID (12), NDIR (11), PID (6), electrochemical cells (6), optical or gravimetric dust measurement (5), liquid-phase parameters — pH, oxidation-reduction potential, and conductivity (4), certified continuous emissions monitoring systems (2), and GC (1). Twenty-eight of the 39 projects required more than one technique. This frequency does not reflect a preference for redundancy; it reflects the structural reality that no single detection principle spans the composition of a real waste gas. An FTIR that quantifies the acid gas fraction reports nothing about total hydrocarbon content; an FID that bounds the hydrocarbon load identifies none of its constituents; and neither technique is sensitive to the particulate phase.

05 Particulate: the distribution decides the technology

For particulate matter, the quantity that determines feasibility is not the one that determines compliance. A permit is written in terms of mass concentration, and the reference method for that value is gravimetric: EN 13284-1 draws a sample isokinetically onto a filter and weighs it. The result is legally robust, and it contains no information whatever about particle size.

Size is what decides whether a wet scrubber can do the job. Collection in a spray tower proceeds by inertial impaction on droplets, and impaction fails as particles become small enough to follow the gas around them. An open spray tower has a cut diameter in the region of 2 to 8 µm; if the mass of the emission sits below that, the technology cannot reach it and no increase in the liquid-to-gas ratio will alter the outcome. Two streams with an identical mass concentration, one coarse and one submicron, therefore, call for entirely different equipment.

Optical aerosol spectrometry supplies what gravimetry does not. By classifying scattered light into size channels — instruments of this class resolve on the order of thirty — it yields the number distribution across the size range, from which the size-resolved mass can be derived. That derivation assumes a particle density and refractive index, so the mass obtained optically is an estimate and not a weighing. The two methods are complementary rather than alternative: the gravimetric value establishes the load and satisfies the regulator, the optical distribution establishes whether a wet scrubber is the right instrument and, if so, what cut diameter it must achieve.

Omitting the distribution is the most common reason a wet scrubber disappoints in service. The installation meets every specification it was given, and the specification was incomplete.

06 From measured concentration to design case

A survey ends in a report; a design process begins with a specification. The step between them is where surveys are most frequently rendered unusable. Three conversions are required.

• Conversion to a mass load. Measured concentrations are converted to mass flow rates using the simultaneously measured volumetric flow, temperature, absolute pressure, and moisture content, and expressed at explicitly stated reference conditions — conventionally mg/Nm³ on a dry basis and, where the applicable sector legislation requires it, corrected to a reference oxygen concentration. Reference conditions that are not stated are reference conditions that will be misread.
• Selection of a design case. A single value must be selected from a measured distribution, and that selection is an engineering decision rather than a statistical one. It is normally specified as a defined high percentile in conjunction with the observed process maximum, with the averaging period stated explicitly. The difference between a peak concentration averaged over one minute and the same peak averaged over thirty minutes can be substantial, and the choice of averaging period must be documented alongside the value itself.
• Assignment of a design margin. The measurement uncertainty quantified in section 03 belongs visibly within the design margin. A margin chosen to accommodate a known and bounded measurement uncertainty is sound engineering; a margin chosen to compensate for the absence of measurement is insurance — paid for in alloy specification, absorbent inventory, and fan energy over the operational life of the plant.

The output of this exercise should take the form of a structured datasheet: identified compounds with their concentrations and associated uncertainties; volumetric flow at stated reference conditions; temperature and moisture content; particulate mass loading with size distribution; the process states under which each parameter was recorded; and the design case derived from the foregoing, with the reasoning made explicit. That document constitutes what a scrubber can be calculated from. It retains its value beyond the measurement campaign itself — at the tender stage, at commissioning, and at every subsequent modification to the plant.

07 What a survey does not settle

Three limitations should be stated plainly, because a measurement campaign is frequently assigned more authority than it can sustain.

The first is that a survey is a snapshot of the plant as it ran during the campaign. A change of feedstock, of production rate, or of recipe can invalidate the characterisation, and a survey conducted prior to a process modification describes a gas that no longer exists. This is an argument for recording the process state with care, not an argument against measurement.

The second is that every technique finds only what it can see. A compound that falls outside every analytical window applied — infrared-inactive, non-ionisable at the lamp energy used, not eluting within the chromatographic run time — remains absent from the report, and its absence is indistinguishable from a true zero. The defence is redundancy of principle rather than accumulation of redundant instruments: combining a sum parameter with a speciating technique and reconciling the two. Where the carbon balance does not close, a compound is present that has not been identified, and that discrepancy constitutes a finding in its own right.

The third is that a survey measures the gas, not its behaviour in a column. Solubility, reaction kinetics, and the number of transfer units required are drawn from the literature and from calculation, and calculation carries its own assumptions — about equilibrium, about mass transfer coefficients, about the absence of interaction between components in a mixture. Where the stakes justify it, the residual uncertainty is resolved by reproducing the measured composition in a pilot column and testing a scrubber configuration against it directly, which converts the final assumption into an empirical observation. That is a separate exercise, and it presupposes the one described here: the mixture cannot be reproduced before it has been measured.