Information measuring system for training assessors to determine the odour threshold using the dynamic olfactometry method

. This article presents the results of training, selection and control of a group of assessors to determine the odour threshold. General recommendations of the European standard EN 13725: 2003 “Air quality － Determination of odour concentration by dynamic olfactometry” for the training, selection and supervision of a group of assessors were used in this work. A trained assessor team or panel is the odour measuring instrument. The dynamic olfactometry method was used to determine the concentration of odour in a gas sample. N-butanol was chosen as the odour sample. The experimental setup for the selection and training of a group of assessors to determine the odour threshold is a information measuring system. This information measuring system provides high accuracy and reliability of measurements. The authors chose the Scentroid SM 100 olfactometer and the Agilent 7800 A gas chromatography-mass spectrometer with a 5975C mass-selective detector as a mean of measuring in information system. The referenced value of odour threshold for n-butanol was set to 38 ppb for a panel of 8 selected n-butanol assessors. The number of steps in the dilution series was 6. The instrumental dilution range was 11 ppb to 145 ppb.


Introduction
Odour is an organoleptic characteristic perceived by the olfactory organ when certain volatile substances are inhaled. An odorant is a substance that stimulates the olfactory system so that a person can smell it.
The ability of humans to smell is associated with survival processes. Unpleasant and putrid food smells scare people away from the food that is hazardous to their health. The smell of burning alerts of a fire. The odour of odorant supplements in the natural gas is associated with the leaks in the gas distribution system.
Odour must be assessed to control the release of volatile odour substances to maintain human health.
The reliability of odor measurement is a subject of much controversy among scientists, ecologists and lawyers. Odour is measured organoleptically. Subjective evaluation by the assessors can lead to large measurement uncertainties. The odour unit is rather difficult to define. Odour must be evaluated in terms of physical quantity.
Special units for measuring odour already exists in several countries, including the European Union. The odour unit [ouE] is the amount of odorant that elicits (causes) a physiological response from the assessor team, when evaporated in 1 m3 of neutral gas under standard conditions. Sensory panel or simply a panel is a group of assessors selected according to a specific procedure. The odour unit is equivalent to one European Odour Reference Mass. The European Reference Odour Mass is the equivalent to 123 μg of nbutanol. Or it is the concentration of 40 ppb n-butanol in neutral gas [1]. The Odour concentration is the number of European odour units per cubic meter of a gas under standard conditions.
In Europe, the dynamic olfactometry has been standardized as the measuring method of odour concentration, and it has been reported that the threshold of n-butanol measured by this method was approximately 40 ppb.
Following the recommendations of the European standard EN 13725: 2003 [1] and applying precision instrumentation techniques could reduce uncertainty in odour changes. This paper presents the odour threshold measurement results using the information measuring system and a group of assessors.
The purpose of the work is to form the panel with approximately the same sensitivity threshold using the information measuring system to control the concentrations supplied to the assessor.

Literature Review
The foul odour from solid waste landfills leads to public complaints and health issues. Several examples are given in publications on the impact of unpleasant odour on public health [2][3][4][5][6][7][8].
Unpleasant odours affect the quality of life and cause various somatic and emotional disorders in people living near scent sources.
Residents of towns who lived near the sources of odours from the coffee factory, bone flour factory, sugar mill, solid household waste dumps were questioned about their health.
The distances from the original source of the odour ranged from 1500 m to 5 km. The analysis showed that odour was the priority factor affecting the health and quality of life of residents.
The intensity and irritation of specific odours in the atmospheric air decrease as respondents move away from the sources. The irritation rate of the unpleasant odours decreases from 87% to 51% depending on the distance from the source from 1500 m to 5 km accordingly. A certain category of residents tended to associate health problems with the environment and mainly with the presence of industrial odours.
This group represents about a third of the population. This group of respondents has a lower tolerance to odours. These people are more likely to suffer from the diseases of the upper respiratory tract, allergies and cardiovascular diseases.
The article [9] presents the results of studies evaluating the application of the University of Pennsylvania Smell Identification Test (UPSIT) in Thailand. Phenyl ethyl alcohol was used as an odour threshold test substance. One hundred and fifty participants took part in this study. The odour threshold was determined in points.
The study was conducted from May 2019 to March 2020. The healthy volunteers took part in the research without any complaints about ability to smell. All participants were tested for odor exposure threshold and odour detection. The overall UPSIT GPA was 26±7 and ranged from 9 to 40. The average lowest odour detection threshold was 6±2.
More than 70 per cent of the test items included fish sauce, banana, coffee, patchouli water, coconut, lemon, orange, our gluten alcohol, vinegar, tea leaf, Thai perfume, jasmine, curry, lime, durian, cola, corn, pineapple, strawberries and grapes.
In the article [10], it is proposed to use not only phenyl ethyl alcohol, but also butanol as a test substance. The authors of the article [11] used n-butanol to determine the odour threshold. In articles [12,13], the authors proposed to use sets of aroma strips of different intensities for the quantitative odour measurement. The intensity of the odour was determined in points on a ranking scale. In the рареr [14], the authors used a gas chromatography-mass spectrometer to measure the odorant concentration.
The article [15] contains data of the long-term studies on the definition of scent of 223 substances. Triangle odour bag method and various instrumental methods were used. Isoamylmercaptan showed the lowest threshold (0.77 ppb) and propane showed the highest threshold (1500 ppm). However, the average odour threshold for all substances was 38 ppb. This corresponded quite well with the pore odour of 40 ppb on n-butanol in [1].
Instrumental methods do not allow detecting low values of some smelling substances, in contrast to sensory methods.
Sensory analysis is an analysis with the help of the senses of the highly specific receptor organs of a person, which provide the body with obtaining information through sight, hearing, smell, taste, touch.
Such modern instrumental methods as electronic noses, mass-spectrometry, chromatography have a lower sensitivity to some volatile odoriferous substances in comparison with the human sensory system.
Long-term research Nagata, Y. [15] by measuring the detection thresholds of chemical substances by instrumental and sensory ones showed that a person senses mercaptan compounds at a level of 1 ppt. While one of the most sensitive detectors, a time-of-flight mass spectrometer, detects these compounds two orders of magnitude higher.
Consequently, the sensory system of humans can detect smelling chemicals in cases where instrumental methods do not detect them.
In the paper [16], the authors used olfactometer for diluting the gas sample. These are examples of the application of instrumental techniques to accurately measure odorant concentrations.

Experimental Methods
General recommendations of the European standard EN 13725: 2003 [1] for the training, selection and supervision of a group of assessors were used in this work.
Odour researchers have not used routinely trained assessors in the past. The physiological response to the number of dose dilutions of the measured sample was determined approximately. The method of standardizing the group of assessors is currently being used. The assessors are selected to the panel by choosing members of the panel with a known sensitivity to the accepted reference material. EN 13725: 2003 suggests the use of n-butanol as a reference odour sample. Assessors must be 16 years of age or older and ready to follow instructions.
A trained assessor team or panel is the odour measuring instrument. Direct quantitative measurement of the concentration of odour in the air is measured using the olfactometry method. Olfactometry is the measurement of the response of assessors to an olfactory stimulus.
The threshold value for the concentration of the odorant is used to assess odour. The detection threshold is the lowest concentration of odorant found by the 50 % of the team of assessors. The assessors should not be aware of the sample's concentration. The threshold concentration of the odorant is created by a series of dilutions with neutral gas. An olfactometer is used to dilute the odorant with a neutral gas at a certain ratio.
EN 13725: 2003 does not provide a precise indication of the type of measuring instrument for the selection of panel members. This document suggests using an olfactometer to dilute the odorant and measuring instrument to measure concentrations. Olfactometer is an instrument in which a sample of odorized gas is diluted with a neutral gas in a specified respect and provided to assessors. The olfactometer must be calibrated using certified gas samples and measuring instrument. The authors chose the Scentroid SM 100 olfactometer and the Agilent 7800 A gas chromatography-mass spectrometer with a 5975C mass-selective detector as a mean of measuring.
The experimental setup for the selection and training of a group of assessors to determine the odour threshold is the information measuring system. This system complies with the requirements of EN 13725: 2003. The information measuring system provides high accuracy and reliability of measurements.
The selection and training of assessors was carried out using the Scentroid SM 100 olfactometer shown in Figure 1. A certified sample of n-butanol in nitrogen with a concentration of 0.989 ppm and an expanded uncertainty of 1.2 % was used as a reference sample. The dynamic method was used to dilute the n-butanol standard sample. High purity nitrogen was used as a neutral diluent gas.
The inlet port, dilution plate, dilution adjuster, and dilution marks on the olfactometer are shown in Fig. 1. The dilution ratio was adjusted using a sliding valve called an odour regulator. The dilution factor was gradually decreased until the assessor detected an odour, at which time the measurement was stopped. The dilution position on the dilution regulator corresponds to the desired concentration of n-butanol. The SM 100 Olfactometer provides 15 dilution levels to allow the required instrumental dilution range. The instrumental dilution range was 11 ppb to 145 ppb. The leader of the group managed the dilution of the sample on the olfactometer. The group leader asked panel members whether they felt the smell or not.
The necessary accuracy of the experiments was ensured by using information about the concentration of n-butanol provided to the assessor. This information was only known to the panel leader. The olfactometer was calibrated against n-butanol using an Agilent 7800 A gas chromatography-mass spectrometer with a 5975C mass-selective detector. A certified standard sample of n-butanol with a concentration of 10 ppm in nitrogen with an expanded uncertainty of 1.2% was used to calibrate the olfactometer.
Calibration of any analytic instrument is an essential condition for obtaining reliable measurement results. The odour measuring devices are also subject to periodic calibration. Annual sampling is a prerequisite for obtaining the required dilution ratio.
Calibration of an olfactometer is usually performed by carbon monoxide. It is a stable gas and it is easy to measure with simple instruments. N-butanol was used to calibrate the olfactometer in this work because it was the substance that was used by the test subjects. The measurement of n-butanol concentration using the gas chromatography-mass spectrometer with a mass-selective detector gives a more reliable result.
The calibration was carried out as follows: pure nitrogen with was supplied to the olfactometer from the main line. The n-butanol sample bag was connected to the inlet port of the cylinder with a certified n-butanol sample. The sample was taken at a distance of 2-5 cm from the sample port. Subsequent analysis was performed on an Agilent 7800 A gas chromatography-mass spectrometer with a 5975C mass-selective detector. The gas flow rate was 40 ml per minute. Concentration of n-butanol measured on Agilent 7800 A gas chromatograph using selected ions at 230 °C. The results were recorded using a mass number of 56, corresponded to n-butanol. The measurement results were processed using an application software.
EN 13725: 2003 recommends calibration once a year. Calibration was performed every three months in this study. The maximum total uncertainty of the measurement results was no more than 5%.
The installation diagram is shown in Fig. 2. The same setup was used to select assessors. The assessor was located at the same place from which the sample was taken when calibrated the olfactometer. The correct answer that the assessor sensed the n-butanol smell in the sample with a threshold concentration was the criterion for selecting him/her (the assessor) to the panel. The sample was given to the assessor with the sample port. The assessor breathed air 2-5 cm away from the sample port. The panel leader set the appropriate dilution factor on the olfactometer. A sample with a known exact concentration of n-butanol was presented to the assessor. Only the panel leader knew this concentration. The sample port served as the interface between the assessor and the installation. This is shown in Fig. 3.  Fig. 3. Interface between assessor and olfactometer.
All experiments on the assessors selection were carried out in a special room. Each assessor entered the room alone. The room has been protected from noise, vibration and other distractions. The room was free of all extraneous odours.
The experimental determination of the individual odour threshold for n-butanol was performed for each assessor candidate in panel. Half of the candidates in the panel were rejected during the experiment. The reasons for rejection were different: inconsistent sensitivity to odours, inability to participate in the experiment for an extended period time. Olfactory responses had to be constant throughout the day and from day to day. Based on the experimental results, the panel was formed from the assessors suitable for determining the odour threshold.
The group of candidates in the panel initially consisted of 15 people (9 men and 6 women). All candidates were familiarized with the assessor code. Each candidate received a unique identification code.
The results were processed in accordance with the algorithm described below. The geometric mean of the odour threshold estimates should have fallen in the range from 0.5 to 2 times the referenced n-butanol concentration. The referenced value for n-butanol was set to 40 ppb. The number of steps in the dilution series was 6. The instrumental dilution range was 11 ppb to 145 ppb. The geometric mean of the individual odour threshold in accordance with (1) is: where Z1 and Z2 are the concentrations corresponding to the extreme concentrations concentration of n-butanol, ZITE is an individual threshold estimate. At least ten measurements had to be made for each candidate to the panel. Two people were unable to participate in all measurements. Therefore, the number of candidates was reduced to 13 at this stage.
The mean of the results and the standard deviation according to formula (2) were calculated using experimental estimates in a logarithmic representation to select panel members: where SITE is the standard deviation for individual odour ratings of each assessor; ZITEi -individual assessment of odour by the assessor in dimension i; ̅̅̅̅̅̅ is the average value of the test results for the assessor. All values were checked for compliance with the selection criteria: -the standard deviation must meet the requirement 10 < 2.3 -the average value must fall within the range: 0.02 < 10 ̅̅̅̅̅̅ < 0.08. (3)

Results and Discussion
8 people out of 15 met all selection criteria. The results that the assessors showed that met both selection criteria are presented in Table 1 in the second and third columns. The first column contains the code of the assessor who was selected for the panel. Column 4 presents the odor threshold results for n-butanol for selected assessors. Разница в абсолютных значениях концентраций н-бутанола меняется от 20 до 61 ppb. This difference seems to be big. However, human perceptions are evaluated on a logarithmic scale. In this case, the values change slightly from 1.3 to 17. The average individual odor threshold for all assessors was 38 ppb. This is close to the reference odor threshold of 40 ppb [1].
The panel was formed as the result of the conducted research. It is necessary to carry out control checks of the panel at least 2 times per year to keep the effectively working panel with the acquired skills. The following factors ensure the reliability of the studies to determine the odour threshold: -Implementation of general recommendations EN 13725:2003; -Use of a certified n-butanol sample with a concentration of 10 ppm; -Calibration of the olfactometer against n-butanol; -Use of the Agilent 7800 A chromate-mass spectrometer with the 5975C mass selective detector to calibrate olfactometer and periodic control of the samples supplied to the assessor; -Periodic control of panel members. The panel is an analogue of the measuring system. The panel can be created using the installation presented in the article.
With the help of the installation, the panel leader sends the known n-butanol concentration to the assessor with great precision.
The assessor responds in the form of a response whether he senses a smell or not. This integration represents a cyber-physical system based on a combination of equipment, computing technology and panels with selected testers.
The reliability of the results is achieved by comparing a measuring instrument with a standard through a continuous chain of calibrations. The reference panel on odour threshold is not currently available.
The system presented in this paper could be used to compare different panels, including those from other countries. This could be analogous to interlaboratory comparisons. At present, the system and panel presented in the work are being used to develop methods for determining odour thresholds for various chemicals. This is relevant for compliance with industrial health and safety regulations.
The use of machine learning technologies will allow better selection and training of testers. The group leader determines in which sequence the concentration is given to the tester at present.
These values are based on the experience and intuition of the group leader. Due to a human factor, an error may occur. Using a self-learning program based on neural networks will help minimize error.

Conclusion and Acknowledgments
This work is the first attempt, at least in Russia. The use of neural networks could be the development of a selection method for a group of assessors. The authors express their gratitude to all people who agreed to participate in this experiment.