DOI: 10.3303/CET24112018 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Paper Received: 4 March 2024; Revised: 18 April 2024; Accepted: 28 June 2024 
Please cite this article as: Cangialosi F., Fornaro A., 2024, Advancements in Instrumental Odour Monitoring System from Stationary Emission 
Sources, Chemical Engineering Transactions, 112, 103-108  DOI:10.3303/CET24112018 
  

 CHEMICAL ENGINEERING TRANSACTIONS  
 

VOL. 112, 2024 

A publication of 

 

The Italian Association 

of Chemical Engineering 

Online at www.cetjournal.it 

Guest Editors: Selena Sironi, Laura Capelli 

Copyright © 2024, AIDIC Servizi S.r.l. 

ISBN 979-12-81206-13-7; ISSN 2283-9216 

Advancements in Instrumental Odour Monitoring System from 

Stationary Emission Sources 

Federico Cangialosia*, Antonio Fornarob  

aTecnologia e Ambiente (T&A), Putignano (BA), Italia  
bLabservice Analytica, Anzola Dell’Emilia (BO), Italia  

federico.cangialosi@icloud.com 

In the past few years, significant advancements have been made in the development of Instrumental Odour 

Monitoring Systems, though very little data is available for IOMS measurement at emission sources and it is 

believed that the main cause is associated with issues affecting continuous sampling from streams with high 

humidity. In this paper we show an approach developed for continuous sampling from flow ducts aimed at 

feeding a trained IOMS with DO data collected from the same sampling system. The system was devised to 

sequentially sample from two duct flows, enabling the characterization of input and output odour flows through 

an odour abatement device while monitoring its efficiency in odour reduction. The primary design issues 

addressed include: 1) decreasing relative humidity through controlled dilution, 2) implementing hardware and 

software adjustments to account for temperature fluctuations during extended operation and 3) developing 

algorithms to establish correlations between odour concentrations obtained through sampling/measurement 

according to EN 13725:2022 and the instrumental odour concentration provided by IOMS. Here we present the 

field-calibrated results of the continuous sampling coupled with IOMS:  the device was installed across the flow 

ducts both upstream and downstream of an odour abatement unit within a waste treatment plant. After the final 

training/calibration, the system operated for several weeks and the collected data of instrumental odour 

concentration are presented: the analysis covers both the odour fluctuations over the plant’s daily working cycle 

and the efficiency of the installed abatement unit in reducing instrumental odour concentration. 

1. Introduction 

The adoption of more stringent regulatory measures, driven by an increased sensitivity of the population to 

odour emissions, has spurred research and development efforts aimed at monitoring and controlling odour 

emissions from industrial facilities. Over the past decade, Electronic Noses or Instrumental Odour monitoring 

systems (IOMS) have gained importance, due to the adoption of low-cost sensor arrays and the widespread 

capability of machine learning algorithms. This technology has been applied in several environmental case 

studies, especially in situation where detecting odour immissions, i.e. the odours of the ambient air along the 

fenceline of the industrial plant, is a significant concern (Oliva at al., 2021, Cangialosi at al., 2021, MacDondald 

et al. 2022). On the other hand, while emissions are often well-defined in terms of sampling procedures for 

analysing odour concentration, especially for stationary emission sources (as outlined in EN 13725:2022), there 

is little emphasis on continuous monitoring specifically aimed at detecting odour compounds. Consequently, 

there is a notable scarcity of literature data concerning the measurement of Instrumental Odour Monitoring 

Systems (IOMS) at emission sources. In such scenarios, where DO samples are typically available for 

straightforward training and calibration tasks, challenges may arise due to issues associated with continuous 

sampling from streams with high humidity. One noteworthy case study, where IOMS was installed near to the 

emission source (Prudenza e al., 2023), involved the use of a dilution system. In this paper, a technology is 

described for continuous sampling and subsequent analysis of airflows within flow ducts for determining odour 

concentration. The system (OSMOTREFF®) was designed to sequentially sample from two distinct duct flows, 

allowing for characterization of both input (hereafter defined as “raw”) and output (defined “clean”) airflows 

passing through an odour abatement device. This assessment involves measuring the chemical concentration 

103



of selected compounds and subsequently quantifying the Instrumental Odour Unit (IOU/m3) after a 

comprehensive training and calibration process. This calibration is conducted using samples collected from the 

same system and analysed using Dynamic Olfactometry (DO) according to EN 13725:2022. 

2. Materials and methods 

2.1 Odour emission continuous sampling: prototype development  

In Figure 1 the schematic of OSMOTREFF® system is shown (within the dotted square), along with the sampling 

lines from an industrial odour abatement unit.  

 

Figure 1: OSMOTREFF® schematic (within the dotted square), along with the sampling lines from an industrial 

odour abatement unit 

The main design challenge tackled was reducing relative humidity through controlled dilution: in fact, finding 

stack emissions with humidity levels below 80% is highly unusual, particularly in emissions associated with 

waste or wastewater treatment plants. Under such conditions, sampling lines quickly become clogged with 

condensed water, while sensor corrosion or damage is likely to occur within a few weeks of continuous 

operation. To address this primary concern, a dynamic dilution system was devised by using two diaphragm 

pumps, operating in a master/slave configuration and digitally controlled by a PLC. Diaphragm pumps were 

selected because of their characteristics: high inertia, minimal overheating, and the absence of lubricated parts 

in contact with the sample. These features mitigate issues such as memory effect, contamination of the pumping 

line, or alterations in the odour profile of the gas flow. One of the two pumps (master pump) is connected to a 

treatment system capable of removing moisture and odour compounds from dilution ambient air. The fluidic 

design of the OSMOTREFF® system incorporates a dilution chamber constructed from inert material. Within 

this chamber, the dilution air flow is regulated based on user-defined settings. Subsequently, the auxiliary pump 

draws the mixed gas flow from the chamber, directing it to the sensor chamber. The flow rate of the auxiliary 

pump, is always equal to or greater than that of the master pump supplying dilution air. If Q0 is the flowrate 

drawn from the sampling line (whether “raw” or “clean”) with a concentration C0 and αQ0 represents the dilution 

air flowrate, then the dilution rate Dil can be defined as follows: 

𝐷𝑖𝑙 =
𝛼𝑄𝑜

𝑄𝑜 + 𝛼𝑄𝑜
=

𝛼

1 + 𝛼
 (1) 

where the denominator is the flow rate of the auxiliary pump extracting the mixed gas from the chamber.   

Given the critical role of the diaphragm pumps in the fluidic setup of the dynamic diluter, both pumps are 

precisely regulated by two mass flow controllers equipped with integrated solenoid valves to ensure optimal 

control. To monitor potential condensation within the diluted stream, temperature and humidity sensors within 

the dilution chamber continuously provide digital readings to the controller. Within the PLC, an operational 

routine has been implemented to oversee the different sampling phases from the "raw" and "clean" streams, 

with intermittent fluidic and sensor chamber rinsing cycles. As a result, the system operates in a continuous 

loop, executing a cyclical program. This program entails sampling from either the "raw" or "clean" stream at the 

designated dilution ratio, with each interval interposed by a rinsing phase. Consequently, a complete cycle 

104



comprises the following sequence: raw-rinse-clean-rinse. During the rinsing phase, only filtered ambient air is 

introduced into the system to cleanse the lines and sensor chamber after each sampling step, thus preventing 

any potential cross-contamination between the two streams. The PLC is also programmed to control a group of 

actuated solenoid valves within the flow distribution manifold to distribute the mixed gas from the dilution 

chamber onto the following lines: 

a) IOMS: Dedicated line to sensors with a flow rate adjustable from 1 L/min up to a maximum of 3 L/min. 

b) Bag 1&2. Dedicated lines for filling bags for olfactometric analysis according to EN 13725:2022. 

c) Sorbent Trap. Dedicated line for sampling adsorbent supports for chemical analysis in the laboratory.  

2.2 Integrating the diluter with IOMS: odour continuous sampling and measuring  

The IOMS (NetPID Plus, Labservice Analytica, Anzola Dell’Emilia (BO), IT) sensor chamber, made of inert and 

odourless material, is fluidically interfaced onto the gas line downstream of the auxiliary pump of the dilution 

module trough the flow distribution manifold. The sensors installed in the IOMS chamber include a 

photoionization detector (PID) calibrated in isobutylene for VOCs, and two electrochemical sensors for H2S and 

NH3. However, the model employed in this case study was equipped with only VOC and H2S sensors. The 

chamber's sealing integrity ensures both measurement accuracy and the protection of electronic components, 

while also preventing potentially hazardous gas accumulations inside the box due to gas leaks. After 

measurement, the gas is exhausted from the system box through a forced air circulation system, which was 

designed also to mitigate the impact of temperature fluctuations on sensor response, thus enhancing overall 

performance and reliability. The data acquisition and processing system operates in two distinct logical phases: 

A. Initially, it computes incremental concentration values relative to background conditions for each 

sensor (C(t)- CBl), using the raw data acquired, which contains aggregated information from processing 

cycles (dilution ratios for “raw” and “clean” stream, sampling and rinsing time). 

With reference to the schematic in Figure 1 and bearing in mind the definition provided in Eq (1), assuming the 

dilution chamber is completely mixed, it is simple to establish that the concentration C(t), for each chemical 

compound detected in the sensor chamber, follows a first-order ordinary differential equation (ODE), the solution 

of which is: 

(𝐶(𝑡) − 𝐶𝐵𝑙)

(𝐶𝑚𝑖𝑥 − 𝐶𝐵𝑙)
= (1 − 𝑒−𝑡/𝜏) (2) 

where: 

CBl is the concentration baseline, when only filtered dilution air is supplied to the system; 

Cmix indicates the concentration achieved at equilibrium when the feed stream (either “raw” or “clean”) with a 

concentration C0 and dilution stream (flowrate αQ0) are mixed, equal to C0/(1+α); 

  is the time constant of the dilution process, given by the ratio between the volume of the dilution chamber 

(Vmix) and the total flowrate resulting from the dilution process: Qmix = Q0(1+α). 

In the typical range of operation, Qmix is around 2 l/min and  is Vmix/Qmix =4.8 s. Hence, the time (t90 mix) one 

must wait to have 90% of the final mixed concentration Cmix is 2.3  = 11 s. A typical EC sensor (such as for 

H2S) has a step response time (t90 sensor) of 40-50 seconds. This suggests that the mixing/diluting phase is not 

the rate-limiting step in the process of sampling-diluting-measuring, ensuring that the signal measured is not 

affected by the continuous sampling device.  

B. In the second phase, an algorithm, trained on DO samples collected by the sampling system itself, 

utilizes the incremental concentration data to calculate the instrumental odour units (IOU/m3) for both 

the “raw” and “clean” streams. 

The instrumental odour units data for the two sampling streams and the efficiency of odour reduction are 

synchronized on a cloud platform, enabling real-time access to the data, trend analysis based on user defined 

time interval, and the setting of threshold values for automatic email notifications in case of exceeding the set 

conditions. 

2.3 Plant installation and on-site training 

The data presented in this present paper are derived from the first long-duration experimental campaign, which 

lasted several months. OSMOTREFF® was installed in a MSW treatment plant to evaluate the performance of 

a wet scrubber treating the exhausted air (25,000 Nm3/h) before a biofilter. The biofilter in the study is an active 

area source, so it was not possible to install OSMOTREFF® to sample the air after biofilter treatment and 

evaluate biofilter’s efficiency, but this should be feasible in the case of a closed biofilter, where the treated air is 

emitted from a stack. Following several preliminary tests and system improvements, particularly addressing the 

temperature issue initially affecting sensor response, the system underwent optimization in terms of cycle 

duration (raw-rinse-clean-rinse) and dilution ratio Dil to achieve optimal sensor response. In fact, as mentioned 

105



in the previous paragraph, the duration of the "raw" (or "clean") sampling phase and the subsequent rinsing 

phase is not constrained by the sampling/dilution phase, but rather determined by the shortest sensor response 

time. For the present case study, each cycle was configured to last 5 minutes, allowing for the measurement of 

three data points from the "raw" stream (untreated air sample) and three data points from the "clean" stream 

(air sample after wet scrubber treatment) within 1 hour. With this setup, it is feasible to collect up to six sample 

bags per hour for Dynamic Olfactometry (DO) analysis in accordance with EN 13725:2022 (three for the "raw" 

stream and three for the "clean" stream). For the development and training of the final algorithm used to derive 

instrumental odour unit from the sensors' response, a total of 42 samples were collected and analysed using 

Dynamic Olfactometry (DO) according to EN 13725:2022 by T&A laboratory. The algorithms used for 

instrumental odour calculation are described elsewhere (Cangialosi et al., 2021); here, it is sufficient to mention 

that fewer signals and features can be extracted, analysed, and utilized for machine learning algorithms due to 

the low number of sensors. However, this is compensated by the clearer sensors’ response obtained from 

stationary source emissions, eliminating the need for specific signal pre-treatment or filtering. 

3. Results and discussion  

3.1 Data acquisition  

In the following figure the raw data acquired by PID and electro chemical sensors are shown. The duty cycles 

are easily recognizable as the higher peaks (for VOC and H2S) correspond to the analysis of the raw gas, the 

lower peaks correspond to the analysis of the clean gas, while the rinsing phases in between a “raw” and “clean” 

cycle correspond to the decreasing concentrations pieces of the signal. 

 

Figure 2: OSMOTREFF® raw data response: in blue (left axis) VOC concentration, in red (right axis) H2S 

concentration 

The plots in the first column (Figures 3a and 3b) display the data from the PID sensor (VOC) and the H2S sensor, 

respectively, revealing a trend with periodic spikes representing the sensors' responses to the “raw” and “clean” 

streams. Once the durations of each phase of the duty cycle (raw-rinse-clean-rinse) are established, a data 

processing routine calculates the incremental concentration values relative to background conditions for each 

sensor (C(t)-CBl) and for each stream. These incremental concentration values are depicted in Figures 3c and 

3d for VOC, where data collected from the untreated samples are plotted in continuous (red) lines. Figures 3e 

and 3f illustrate the H2S data for both the "raw" and "clean" streams. By simply examining the scales of the plots, 

it can be pointed out that the scrubber can achieve approximately a 50% reduction in terms of VOC. However, 

H2S does not appear to be affected by the wet scrubber treatment, as expected, given that the scrubber does 

not employ chemicals (like NaOH) to treat reduced sulphur compounds.  

In Figure 3, the monitoring results of a 24-hour scrubber operation are presented.  

 

106



 

  

Figure 3a-f: OSMOTREFF® data: Figures a) and b): raw data responses for VOC and H2S.  Figures 3c and 3d: 

VOC incremental concentrations for the “raw “and “clean” streams. Figures 3e and 3f: H2S incremental 

concentrations for the “raw “and “clean” streams  

3.2 Instrumental Odour Unit from sensors data  

In Figure 4, the Instrumental odour unit for the same 24-hour time interval as in Figures 3a-f is displayed.  

 

Figure 4: OSMOTREFF® data response in IOU/m3 (in blue clean concentration, in red raw concentration)  

A detailed discussion regarding the correlation between DO analyses and IOMS data will be provided in a 

forthcoming paper. One conclusion worth emphasizing here is that simple single-sensor correlation is not 

sufficient for odour prediction as also recently highlighted by Prudenza et al. (2023). Indeed, it was discovered 

that within the same range of VOC concentrations detected by the PID, different values of odour concentrations 

are observed for the "raw" and "clean" streams, with this difference appearing to increase as the concentration 

rises. Evidently, the complex chemical composition of the "clean" gas differs from that of the "raw" gas, and 

even if the PID sensor, calibrated in isobutylene, yields similar responses for both streams, the corresponding 

odour concentrations can be, and indeed are, noticeably different due to the varying chemical compositions 

before and after treatment. These findings highlighted what follows: simpler and cost-effective multi-sensor 

systems, compared to more complex IOMS used for ambient air or fenceline analysis, can be utilized for high-

concentration applications such as source emissions. However, there is a trade-off: multi-sensors trained as 

IOMS for stationary emission sources may lack generality, requiring training to be conducted considering "clean" 

data separate from "raw" data. Additionally, the training algorithm should be selected to account for all sensor 

responses, exploring nonlinear correlations with DO data. Chemical mixtures containing two or more chemicals, 

a)

0

0.2

0.4

0.6

0.8

1

1.2

0
0
:0
9
:4
3

0
1
:0
9
:4
7

0
2
:1
0
:5
3

0
3
:1
1
:2
9

0
4
:1
2
:0
5

0
5
:1
2
:1
1

0
6
:1
1
:1
7

0
7
:1
0
:2
4

0
8
:1
0
:5
8

0
9
:1
1
:3
4

1
0
:1
2
:1
0

1
1
:1
2
:4
7

1
2
:1
1
:5
3

1
3
:1
0
:5
8

1
4
:1
1
:0
5

1
5
:1
1
:4
1

1
6
:1
2
:1
7

1
7
:1
2
:5
3

1
8
:1
1
:5
9

1
9
:1
1
:0
5

2
0
:1
1
:1
2

2
1
:1
1
:4
8

2
2
:1
2
:2
4

2
3
:1
2
:5
9

V
O

C
(p
p
m
)

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0
0
:0
1
:0
8

0
1
:0
0
:4
7

0
2
:0
0
:2
2

0
2
:5
9
:5
8

0
3
:4
3
:3
2

0
4
:4
3
:0
8

0
5
:4
1
:1
3

0
6
:2
3
:4
8

0
7
:2
2
:2
5

0
8
:2
1
:5
9

0
9
:2
1
:3
5

1
0
:2
1
:1
1

1
1
:2
0
:1
8

1
2
:0
2
:5
2

1
3
:0
0
:5
7

1
3
:4
3
:3
1

1
4
:4
3
:0
8

1
5
:4
2
:4
3

1
6
:4
2
:2
0

1
7
:4
1
:2
6

1
8
:2
4
:0
1

1
9
:2
2
:0
6

2
0
:2
1
:1
3

2
1
:2
0
:4
8

2
2
:2
0
:2
4

2
3
:0
3
:5
8

Raw Clean

0

20

40

60

80

100

120

140

160

180

0
0
:0
1
:0
8

0
1
:1
9
:4
8

0
2
:2
2
:2
4

0
3
:4
1
:0
2

0
4
:4
3
:3
8

0
6
:0
0
:1
6

0
7
:0
1
:2
2

0
8
:1
9
:5
9

0
9
:2
2
:3
6

1
0
:4
1
:1
4

1
1
:4
2
:5
0

1
2
:4
3
:5
6

1
4
:0
1
:0
3

1
5
:0
3
:4
0

1
6
:2
2
:1
8

1
7
:4
0
:2
6

1
8
:4
1
:3
3

1
9
:4
2
:3
9

2
1
:0
1
:1
7

2
2
:0
3
:5
3

2
3
:2
2
:3
0

H
2
S
(p
p
b
)

0

20

40

60

80

100

120

140

160

0
0
:0
9
:4
3

0
1
:1
0
:1
7

0
2
:1
1
:5
3

0
3
:1
3
:0
0

0
4
:3
0
:0
6

0
5
:3
0
:1
3

0
6
:2
9
:4
9

0
7
:2
9
:5
6

0
8
:3
1
:0
0

0
9
:3
2
:0
6

1
0
:3
3
:1
3

1
1
:3
3
:4
9

1
2
:3
3
:2
6

1
3
:3
3
:0
0

1
4
:5
0
:0
8

1
5
:5
1
:1
5

1
6
:5
2
:2
1

1
7
:5
2
:2
7

1
8
:5
2
:0
3

1
9
:5
2
:1
0

2
0
:5
3
:1
6

2
2
:1
0
:2
4

2
3
:1
1
:2
9

Raw Clean

c) d)

e) f)b)

0

20000

40000

60000

80000

100000

120000

140000

160000

180000

200000

0
0
:0
1
:0
8

0
0
:5
4
:4
6

0
1
:4
8
:2
1

0
2
:4
1
:5
6

0
3
:3
5
:3
2

0
4
:2
8
:3
6

0
5
:2
2
:1
2

0
6
:1
5
:1
7

0
7
:0
8
:2
3

0
8
:0
1
:2
7

0
8
:5
4
:3
3

0
9
:4
7
:3
8

1
0
:4
0
:4
3

1
1
:3
3
:4
9

1
2
:2
7
:2
5

1
3
:2
0
:5
9

1
4
:1
4
:0
5

1
5
:0
7
:4
0

1
6
:0
0
:4
6

1
6
:5
3
:5
1

1
7
:4
6
:5
7

1
8
:4
0
:3
2

1
9
:3
3
:3
8

2
0
:2
7
:1
4

2
1
:2
0
:1
9

2
2
:1
3
:5
4

2
3
:0
6
:5
8

In
st

ru
m

e
n

ta
l O

d
o

u
r 

U
n

it
 (

IO
U

/m
3 )

time

Raw

Clean

107



each with an odour threshold and an Odour Activity Value, often exhibit an odour concentration that differs 

significantly from the linear combination of the activity values of individual chemicals, demonstrating a nonlinear 

and multicomponent behaviour (Szulczyński and Gębicki, 2019). Following an extensive experimental phase, a 

standardized procedure involving 40 to 50 DO samples and an appropriate ML algorithm was developed. The 

validity of the algorithm and the optimal number of training samples were tested by comparing the results of 

odour concentration in DO before and after the treatment with the instrumental odour concentration IOU/m3 

(Cangialosi and Fornaro, 2023). In this case study, 42 samples were employed for ML training and testing. Once 

integrated into the OSMOTREFF® system, the algorithm computes Instrumental Odour Units (IOU/m3) in real-

time based on the incremental concentrations derived from multi-sensor data acquisition.  

It is important to highlight that: 1) "raw" odour concentrations exhibit a fourfold increase within the same day, 

the peaks coinciding with mechanical waste treatment activities in the buildings; 2) the scrubber odour 

abatement efficiency over the 24-hour period is approximately 85%, a remarkably high value, though caution is 

advised in its interpretation as long-term analyses of the system performance may offer more representative 

data. The odour treatment efficiency is significantly higher than that derived from the analysis of single sensor 

concentrations (50% for VOC and nearly zero for H2S). This discrepancy arises because the odour 

concentrations associated with PID values obtained from the "raw" stream are much higher than those obtained 

from the "clean" stream. This highlights the importance of not exclusively using single sensors concentrations 

to evaluate odour reduction efficiency. Without utilizing a comprehensive dataset of DO samples alongside 

properly designed machine learning algorithms, such an approach can result in highly misleading outcomes. 

4. Conclusions 

In the paper, we have summarized the advancements made in the field of instrumental odour monitoring 

systems for stationary emissions. These advancements enable the characterization of input and output airflow 

odour concentrations through a wet scrubber while monitoring its efficiency in odour reduction. The schematic 

of the device (OSMOTREFF®), designed for continuous sampling with dynamic dilution, is presented alongside 

the integration with an IOMS consisting of one PID sensor and two electrochemical sensors for hydrogen 

sulphide and ammonia detection. Considering the typical response time of the sensors, which has been 

identified as the rate-limiting step in the sampling-diluting-analysing process, the system can measure up to 6 

data points per hour, i.e. three data points from the untreated sample line ("raw"), and three from the sample 

line after the wet scrubber ("clean"). After training the IOMS with 42 samples analysed in dynamic olfactometry, 

instrumental Odour unit (IOU/m3) are calculated form the sensors response curves in real time. Field data 

collected over a 24-hour period from an industrial wet scrubber, which treated air from a municipal solid waste 

treatment plant, showed notable variations in odour concentrations on a daily basis, with fluctuations of up to 

400% in terms of peak-to-baseline variation. The scrubber's treatment efficiency for odour abatement was 

approximately 85% during the monitoring period, while single sensor concentration reduction was 50% for VOC 

and negligible for hydrogen sulphide. This difference is explained by experimental evidence indicating that odour 

concentrations (determined according to EN 13725:2022) corresponding to the same PID values measured in 

the "raw" stream are significantly higher than those measured in the "clean" stream. This underlines the 

importance of calculating instrumental odour concentrations following an appropriate on-site training/calibration 

procedure, as single sensors measurements alone may be misleading when analysing odour concentration 

trends. 

References 

Cangialosi F., Bruno E., de Santis G., 2021, Application of Machine Learning for Fenceline Monitoring of Odour 

Classes and Concentrations at a Wastewater Treatment Plant, Sensors, 21, 4716.  

Cangialosi F., Fornaro A., 2023, IOMS in-situ training and validation for monitoring odour abatement efficiency. 

Conference Odour emissions: from reduction technology to new control strategies, November 7th, Rimini, 

IT. (in Italian) 

MacDonald M., Thoma E., George I., Duvall R., 2022, Demonstration of VOC Fenceline Sensors and Canister 

Grab Sampling near Chemical Facilities in Louisville, Kentucky, Sensors, 22, 3480.  

Oliva G., Zarra T., Pittoni G., Senatore V., Galang M.G., Castellani M., Belgiorno V., Naddeo V., 2021, Next-

generation of instrumental odour monitoring system (IOMS) for the gaseous emissions control in complex 

industrial plants, Chemosphere, 271,129768. 

Prudenza S., Bax C., Capelli L., 2023, Implementation of an electronic nose for real -time identification of odour 

emission peaks at a wastewater treatment plant, Heliyon, 9(10), E20437. 

Szulczyński, B., Gębicki, J., 2019, Determination of Odour Intensity of Binary Gas Mixtures Using Perceptual 

Models and an Electronic Nose Combined with Fuzzy Logic, Sensors, 19, 3473.  

108


	5cangialosi.pdf
	Advancements in Instrumental Odour Monitoring System from Stationary Emission Sources