Scientific blog · Research

Removal efficiency for dust and organic compounds in flue-gas treatment: evidence from instrumental stack tests

Artur Cokaj · 6 October 2026 · Updated 21 October 2026

Analysis of stack test reports on three treatment configurations (water scrubber only, filtration, filtration with electrostatic filter): UNI methods, percentage reduction in concentration and mass flow, limitations and regulatory framework.

Abstract

Purpose. To assess removal efficiency for dust and organic compounds in three sets of instrumental stack tests, divided by treatment configuration: water scrubber only (group A), filtration and filtration with an electrostatic filter on a roasting process (groups B and C), and grilling with treatment (group D).

Methods. Isokinetic sampling with a Pitot tube, gravimetric dust measurement (UNI EN 13284-1), total organic carbon measured using a flame ionisation detector (UNI EN 12619), organic speciation (UNI CEN/TS 13649), flow rates according to UNI 10169 and UNI EN ISO 16911-1, and ozone using OSHA method 214. Removal is calculated in terms of concentration and mass flow.

Results. With the water scrubber alone, dust decreased from 516,5 to 24,8 mg/Nm³ (95,2 % by concentration, 93,6 % by mass). With filtration, removal relative to the untreated test was 42-72 % for dust and 73-79 % for COT; the cooling, filter and electrostatic filter configuration yielded 2,3 mg/Nm³ of dust and 10,8 mg/Nm³ of COT. In group D: dust 3,2 mg/Nm³, organic compounds 17 mg/Nm³, ozone <0,05 mg/Nm³.

Conclusions. Treatment measurably reduces both pollutant classes, with differences between configurations. A single sample per configuration and non-simultaneous sampling limit the scope of the conclusions; comparison with regional and local limits depends on the oxygen reference specified in the permit.

Keywords: water scrubber, electrostatic filter, dust, volatile organic compounds, atmospheric emissions, roasting, removal efficiency.

1. Introduction

The efficiency of an emission-control system can only be demonstrated by measuring pollutants before and after treatment using standardised methods and recording the results in a test report. This work brings together three sets of stack test reports produced by external laboratories between 2014 and 2024, arranges them by technology and calculates percentage reduction in concentration and mass flow.

The measurements concern two types of effluent: flue gases from a combustion system treated with a water scrubber, and flue gases from a coffee-roasting oven treated with filters and an electrostatic filter. Tests are identified solely by report number and the technical standards applied; the parties and locations are not identified.

There are four objectives: to present the data with the associated method, calculate removal, discuss the limitations of possible comparisons and place the values within the regulatory framework governing emissions from this activity.

2. Materials and methods

All measurements were carried out at the stack by external testing laboratories, using the UNI EN, UNI and OSHA standards listed in the reports. The table associates each parameter with its method and the groups in which it appears.

2.1 Sampling conditions

Group A. Circular sampling cross-section with a diameter of 0,30 m (0,070 m²), system under normal operating conditions. Two 30-minute samples on the same day (02/12/2014), upstream (14:53-15:23, 0,565 Nm³ sampled) and downstream (15:33-16:03, 0,508 Nm³) of the scrubber.

Groups B and C. Circular cross-section with a diameter of 0,20 m (0,031 m²), sampling on 19/07/2024 at full load (100 %) and constant operating conditions, using an isokinetic probe and S-type Pitot tube. Velocity is measured at three points (insertion depths of 5, 10 and 18 cm). COT is measured continuously using an FID analyser, combustion gases with a dedicated analyser, and organic compounds on thermal-desorption tubes (2-5 L sampled). The four tests cover 11:35-16:10 on the same day. The stated legislative reference is D.Lgs. 152/2006.

Group D. Circular cross-section with a diameter of 0,30 m, sampling on 07/06/2019 from 10:30 to 13:10, under the most demanding operating conditions (grilling at maximum utilisation).

2.2 Roasting test configurations

Report 241972.001 contains a conflict: the heading states “without electrostatic filter”, while the “emission-control system” field specifies “filter + electrostatic filter”. The configuration of this test cannot be assigned with certainty on the basis of incompatible information. The data are presented to identify the documentation issue; the comparison alone does not demonstrate the additional contribution of the electrostatic filter.

2.3 Calculating removal

Removal by concentration is the percentage decrease relative to the upstream value. Removal by mass flow also accounts for the dry normalised flow rate at each point, because a higher downstream flow rate dilutes the concentration without reducing the mass emitted.

For group A, Cin and Cout are the upstream and downstream concentrations. For groups C, Cin is the concentration in the untreated test (241972.003). The concentration corrected to a reference oxygen level is obtained using the following relationship.

Group A reports provide expanded uncertainty for dust (k = 2, 95 % confidence level), used to calculate the worst-case and best-case removal interval. The 2024 reports state that uncertainty is provided on request and assess compliance using the reported value without considering it; the values for groups B and C are therefore presented without an interval.

3. Group A results: water scrubber

Total dust decreases from 516,5 to 24,8 mg/Nm³, with removal of 95,2 % by concentration and 93,6 % by mass flow. Measured values are shown in the table; calculated quantities are identified as such.

Actual flow rate is mean velocity multiplied by the 0,070 m² cross-section. When converted to normal conditions using the temperature, humidity and pressure in the reports, it agrees with the declared dry flow rate within 2 % (approximately 688 versus 700 Nm³/h upstream, 923 versus 930 downstream).

The temperature decrease of 41 °C is consistent with cooling and water washing of the stream. The difference between the two removal results, 1,6 percentage points, depends solely on the higher downstream flow rate measured (+32,9 % in dry Nm³/h, +19,5 % in actual m³/h); possible causes are examined in the discussion.

4. Groups B and C results: roasting without treatment, with filters and with an electrostatic filter

In the numerical comparison with the untreated test, the filter test shows reductions of 42 % for dust and 73 % for COT; report 241972.001 shows reductions of 72 % and 79 %. The configuration of this latter test is uncertain because of the documentation conflict described in section 2.2: the comparison does not demonstrate the effect of adding an electrostatic filter. Tests were conducted on 19/07/2024 at the same stack; the table gives the measured values.

4.1 Removal relative to the untreated test

The arithmetic difference between the reductions calculated for the two tests is 29,4 percentage points for dust, 6,4 for COT and 14,1 for speciated compounds. The conflict over the configuration in report 241972.001 prevents these differences from being attributed to the electrostatic filter. Mass flow for test C2 is calculated using the flow rate of 210 Nm³/h, which is 14 % higher than in tests 241972.002 and 241972.003.

4.2 Organic compound speciation (mg/Nm³)

A dash indicates a compound not listed in the report for that test. In the three roasting tests, the three aldehydes (acetaldehyde, 2-methylpropanal, 2-methylbutanal) and 2-furfurylthiol appear only without treatment. Speciation samples are 2-5 litres: the data identify the compounds present and indicate their order of magnitude.

4.3 Combustion gases and cooling test

CO and NOx are very low in all tests: CO ranges from 1,4 to 4,5 mg/Nm³ and NOx from 0,1 to 0,6 mg/Nm³, with no treatment-related trend. The measured oxygen level, 20,3-20,4 %, indicates a highly diluted effluent, close to ambient air composition. CO2, SOx and acids were not determined.

Test 241972.004 concerns the cooling phase, with a flow rate of 2100 Nm³/h and a velocity of 23 m/s: it is not comparable with the first three. Its concentrations are lower partly because of dilution. COT mass flow (22,7 g/h) is in fact higher than in tests 241972.001 and 241972.002 (7,69 and 8,84 g/h), despite the lower concentration.

5. Group D results: ozone, dust and VOC at the emission point

At the emission point of a fume and odour treatment system, ozone was below the method’s limit of quantification, 0,05 mg/Nm³. Dust and volatile organic compounds were measured in the same sampling exercise, as shown in the table.

The “<” symbol indicates a value below the limit of quantification: the reported mass flow (0,000043 kg/h) is that limit multiplied by the dry flow rate, 0,05 mg/Nm³ × 865 Nm³/h. The result therefore means “not quantifiable”, not “absent”. The limit corresponds to approximately 0,023 ppm under normal conditions (0 °C, 1 atm).

The measurement is a stack analysis using OSHA method 214:1995 at an emission point. It is not an appliance ozone-emission test in a test chamber according to UL 867, which has different geometry, volume and a different limit (0,050 ppm in the test environment).

This group has only one sampling point, downstream of the system: the reports contain no upstream value, so a percentage removal cannot be calculated.

6. Discussion

The data show removal ranging from 42 % to 95 %, but each comparison is limited by how the measurements were organised. This section identifies the causes and their effects on the results.

6.1 Non-simultaneous sampling and flow rate (group A)

Upstream and downstream samples were taken in two successive phases (14:53-15:23 and 15:33-16:03), with a ten-minute interval. Non-simultaneous sampling is the most likely explanation for the difference in flow rate between the two points (700 and 930 dry Nm³/h), because system conditions can change between phases. Supporting this interpretation, the wet-flow uncertainty ranges (±150 and ±200 Nm³/h) overlap: 780 + 150 = 930 Nm³/h upstream and 1060 − 200 = 860 Nm³/h downstream.

A real increase in flow rate would conflict with cooling: at 43 °C instead of 84 °C, the same mass of gas occupies 88,5 % of the volume ((273+43)/(273+84)), whereas velocity increased by 19,5 %. Emitted mass does not increase in either case: 361,6 g/h upstream and 23,1 g/h downstream. The effect of flow rate on the result is limited to 1,6 percentage points (95,2 % by concentration, 93,6 % by mass).

6.2 A non-simultaneous reference in groups B and C as well

The untreated test (241972.003) and treated tests were conducted at different times, between 11:35 and 15:34. Emissions from a roasting oven vary with the cycle phase and batch; the reports do not identify the phase for each sample. The differences between tests therefore include the effect of treatment and, to an unquantifiable extent, process variability. A single sample per configuration prevents repeatability from being estimated.

6.3 Dust, collected mass and uncertainty

Filter mass gains are 0,7-2,5 mg over volumes of 0,05-0,32 Nm³. With such small masses, relative uncertainty for dust is higher than for COT, and dust-removal percentages have a wider margin than the other results. The reports do not provide uncertainty for these tests.

6.4 Reference oxygen

Oxygen measured in the roasting tests is 20,3-20,4 %: the effluent is highly diluted. Applying a 17 % O2 reference as a calculation example gives factors of approximately 5,7 for O2 of 20,3 % and 6,7 for O2 of 20,4 %. These factors differ: when measured oxygen changes, percentages calculated from corrected concentrations must be recalculated. The reference applicable to comparison with an emission limit must be checked in the provision relevant to the installation.

6.5 What the data do not measure

No report contains an olfactometric measurement (UNI EN 13725), dust particle-size distribution, SOx, acids, metals or polycyclic aromatic hydrocarbons. These tests therefore do not demonstrate odour removal or system behaviour over time (filter saturation, cleaning, performance drift). Ozone (group D) was measured at a single point and without an upstream value.

7. Regulatory framework and comparison with limits

7.1 National level

Atmospheric emissions from production facilities are governed by Part Five of D.Lgs. 152/2006. Coffee-roasting facilities with production not exceeding 450 kg/day fall within the derogation activities under Article 272 and Annex IV, and are subject to a general or specific permit. The decree does not establish a single limit for this activity: limits, oxygen references and monitoring frequency are defined by the regions and competent authorities in their own provisions.

7.2 Examples of regional and local limits consulted

The following values were read in the public documents listed in the references and are provided as examples. They do not describe a particular installation or replace the applicable permit.

The provisions listed in the table set different conditions and requirements. The cited monitoring frequencies also differ: the table indicates monitoring every two years for Messina and annually for the Lecce case. These examples do not establish a general monitoring frequency for other activities or installations.

7.3 Indicative comparison with measured values

All test values were measured without oxygen correction. The table also shows them referenced to 17 % O2 using the factor (21−17)/(21−measured O2), namely 5,7 for O2 20,3 % and 6,7 for O2 20,4 %.

Three cautions apply to the comparison.

1. Oxygen reference. Correction to 17 % is meaningful for combustion gases from a direct-fired roaster. Tests in groups B and C measure an effluent with 20,3-20,4 % O2, almost ambient air, and the applicable permit may or may not require correction for that stream. Without correction, values for the cooling configuration (2,3 and 10,8 mg/Nm³) are below all the example limits; with correction, dust for that configuration (13 mg/Nm³) remains close to the lower limits and COT (62 mg/Nm³) exceeds 50 mg/Nm³.
2. Sampling point. A limit applies to the stack identified in the permit. The tests described here are sampling exercises to assess removal, and their results do not amount to an installation compliance check.
3. A single sample per configuration. A compliance check generally requires several samples under representative conditions and comparison taking measurement uncertainty into account.

This study therefore does not declare compliance with the limits. It observes that values for the cooling and filter + electrostatic filter configuration are much lower than those in the other groups, and that the final comparison depends on the oxygen reference chosen by the competent authority.

7.4 Ozone

The documents consulted contain no stack ozone-emission limit for these activities. The group D measurement (<0,05 mg/Nm³, the method’s limit of quantification) is an emissions result. The 0,050 ppm value in UL 867 refers to a different type of test, performed on the product in a chamber, and must not be compared with a stack measurement.

8. Conclusions

The instrumental stack tests analysed in this study show that flue-gas treatment measurably reduces dust and organic compounds, and that the extent of reduction depends on the configuration.

With the water scrubber alone (group A), dust concentration decreased from 516,5 to 24,8 mg/Nm³, with removal of 95,2 % by concentration and 93,6 % by mass flow. With filtration configurations (groups B and C), dust removal relative to the untreated test ranges from 42 % to 72 % by concentration, and total organic compound removal from 73 % to 79 %; the cooling, filter and electrostatic filter configuration yielded the lowest absolute values (dust 2,3 mg/Nm³, COT 10,8 mg/Nm³). In the grilling test (group D), dust and organic compounds were 3,2 and 17 mg/Nm³ respectively, and ozone remained below the limit of quantification (<0,05 mg/Nm³).

The results must be read with the limitations stated in section 6: a single sample per configuration, non-simultaneous upstream and downstream sampling, no dust uncertainty for groups B and C, and no olfactometric or particle-size measurement. Comparison with regional and local limits depends on the oxygen reference specified by the permit, and this study does not declare any installation compliant.

Campaigns with simultaneous, repeated sampling, recording of the roasting-cycle phase and olfactometric measurements are needed to confirm the results and quantify the effect of individual filtration stages.

9. References

Technical standards

UNI 10169:2001, Emission measurements - Determination of velocity and flow rate in ducted streams.

UNI EN ISO 16911-1:2013, Stationary source emissions - Manual and automatic determination of velocity and volume flow rate in ducts. Part 1: Manual reference method.

UNI EN 13284-1:2003 and UNI EN 13284-1:2017, Stationary source emissions - Determination of low-range mass concentration of dust. Part 1: Manual gravimetric method.

UNI EN 12619:2013, Stationary source emissions - Determination of the mass concentration of total gaseous organic carbon at low concentrations.

UNI CEN/TS 13649:2015, Stationary source emissions - Determination of the mass concentration of individual gaseous organic compounds.

UNI EN 14792:2017, Stationary source emissions - Determination of the mass concentration of nitrogen oxides (NOx).

UNI EN 15058:2017, Stationary source emissions - Determination of the mass concentration of carbon monoxide (CO).

UNI EN 14789:2017, Stationary source emissions - Determination of the volume concentration of oxygen (O2).

UNI EN 14790:2006, Stationary source emissions - Determination of water vapour in ducts.

OSHA Method ID-214:1995, Ozone in workplace atmospheres (reference method for ozone).

UL 867, Standard for Electrostatic Air Cleaners (0,050 ppm ozone limit in a test chamber).

Legislation and administrative provisions

D.Lgs. 3 April 2006, No. 152, Part Five (Articles 267-298), specifically Article 272 and Annex IV.

Metropolitan City of Messina, Coffee and other product roasting, 2020 general permit.

Calabria Region, Technical annexes, Decree No. 6852 of 02/07/2021, Sheet No. 9 Coffee roasting.

City of Lecce, Issue of Single Environmental Authorisation (AUA), 2016 administrative determination.