The long and winding road of CEN/TC264/WG41 developing a standard for validating Instrumental Odour Measurement Systems

Ton WG41  In May 2015 the decision was made, by a positive vote in Technical Committee 264 ‘Air Quality” of the Committee Européen de Normalisation (CEN), to establish a preliminary work item “Air quality – Continuous instrumental odorant monitoring in air to assess risks of odour (nuisance) and safety” which was entrusted to working group WG41 "Electronic sensors for odorant monitoring" for a period of 6 years starting on 2015-05-12. The Dutch member of CEN, NEN, was assigned the secretariat and the author of this article was appointed as convenor.

  The first meeting of WG41 was held on the 22nd of October 2015 in Antwerp. After 20 meetings over 6 years, we have come to the end of the time assigned without presenting a final draft standard for voting to the TC 264 “Air Quality.

Anton P. van Harreveld

   Convenor CEN/TC264/WG41 & founder Odournet group ton@odournet.com

 

   Competing interests: The author has declared that no competing interests exist.

   Academic editor:  Carlos N. Diaz

   Content quality: This paper has been peer-reviewed by at least two reviewers. See scientific committee here.

   CitationDr. Anton P. van Harreveld, 2021, The long and winding road of CEN/TC264/WG41 developing a standard for validating Instrumental Odour Measurement Systems. 9th IWA Odour& VOC/Air Emission Conference, Bilbao, Spain, www.olores.org.

   Copyright:  2021 Olores.org. Open Content  Creative Commons license. It is allowed to download, reuse, reprint, modify, distribute, and / or copy articles in olores.org website, as long as the original authors and source are cited. No permission is required from the authors or the publishers.

   ISBN: 978-84-09-37032-0

   Keyword: machine olfaction, e-nose, validation.

Download Presentation

Video

 

Abstract

   In May 2015 the decision was made, by a positive vote in Technical Committee 264 ‘Air Quality” of the Committee Européen de Normalisation (CEN), to establish a preliminary work item “Air quality – Continuous instrumental odorant monitoring in air to assess risks of odour (nuisance) and safety” which was entrusted to working group WG41 "Electronic sensors for odorant monitoring" for a period of 6 years starting on 2015-05-12. The Dutch member of CEN, NEN, was assigned the secretariat and the author of this article was appointed as convenor.

   The first meeting of WG41 was held on the 22nd of October 2015 in Antwerp. After 20 meetings over 6 years, we have come to the end of the time assigned without presenting a final draft standard for voting to the TC 264 “Air Quality.

   The work item will therefore be suspended. Work by members of WG41 can continue in an informal setting, until there is sufficient progress to restart the formal CEN standardisation process with the establishment of a new preliminary Work Item, with a new timeline.

   Despite this unsatisfactory lack of tangible result a significant amount of work has been done, in developing the conceptual framework for the validation of results obtained by Instrumental Odour Monitoring devices (IOMS) for air quality monitoring applications.

   At least the term IOMS has been coined, embedded in a framework of terms and definitions to discuss machine olfaction as distinct from machine olfaction with semantic precision. We now have the terminology to describe and discuss the overlaps and interface between these two domains of chemical detection, the sensory- and the machine olfaction. The difference between odour concentration (ouE·m-3) and the odour stimulus indicator value (OSI) is now crystal clear.

   This paper describes the progress made and the gaps that still need to be filled to prepare a draft standard for validating the results of IOMS monitoring results for air quality management purposes.

 

1. Introduction

   The term “electronic nose” was coined in 1988 by Julian Gardner and Philip Bartlett (Gardner, Bartlett, 1994). The ambition to study olfaction scientifically and quantitatively goes back another century. The first textbook on olfaction was published in 1895 (Zwaardemaker, 1895). Alexander Graham Bell not only proposed the decibel to express signal strength as a multiple of the detection threshold, but also speculated on the numerical approach to measure smell. In 1914 he wrote (Bell, 1914):

   Did you ever measure a smell? Can you tell whether one smell is just twice strong as another? Can you measure the difference between two kinds of smell and another? It is very obvious that we have very many different kinds of smells, all the way from the odour of violets and roses up to asafetida. But until you can measure their likeness and differences, you can have no science of odour. If you are ambitious to find a new science, measure a smell.

   What is an odor? Is it an emanation of material particles into the air, or is it a form of vibration like sound? If you can decide that, it might be the starting point for a new investigation. If it is an emanation, you might be able to weigh it; and if it is a vibration, you should be able to reflect it from a mirror.

Alexander Graham Bell, 1914

   And “to measure a smell” is precisely the ambition we have chased in past decades, in the context of air quality management and regulation. For sensory quantification of odours technical standards were developed since the 1980’s (Schulz, van Harreveld, 1996) culminating in the reference method for measuring odour concentration in the CEN standard EN13725, first published in 2003, now revised in 2021. However, for machine olfaction, there are still gaps to be filled before IOMS devices can fulfil the promise of continuous odorant gas monitoring for air quality management purposes.

   The demand for IOMS devices for air quality monitoring has been in evidence, and in the application of IoT to environmental control IOMS devices and platforms to visualise the results are a common feature. However, the main question remains: What is it that an e-nose detects? And what does it say about our human sensory experience of odour?

   The precondition for application of IOMS devices for air quality monitoring is our ability to validate the degree in which the measured indicator value of an IOMS is predictive for the human olfactory sensation and the formalised sensory measurement of smell, as defined in EN13725 and EN16841. Designing this validation mechanism was the main challenge for the working group CEN/TC264/WG41 for the past 6 years, and the job is not yet finished.

   The scope of work for WG41 was defined very specifically to include the following measurement tasks where an IOMS could be used:

  • Detecting the absence or presence of the odour under study
  • Classifying the presence of multiple odours under study (odour A, odour B, ….., odour X) or the absence of all of these odours
  • Measuring the odour stimulus indicator, as a measure of the amount of odour present, an indicator for odour concentration.

   IOMS can be applied at the source, at the fence line or at a receptor location.

 

2. Approach and progress

   The task of WG41 has turned out to be a complicated one. As a first step the scope of work needed to be delineated.

   WG41 took an early decision to not consider the IOMS instruments and their technical design or workings. The variety of instruments on the market, the variety of sensors and the variety of data processing approaches was significant. Therefore, the task of finding a technical standard for the instruments was considered too challenging, also considering that the field is still under development. Neither did WG41 consider the training of these instruments, which uses a variety of methodologies, from simple linear approaches combining a few sensor signals to advanced AI algorithms. As a consequence, WG41 considers the IOMS as a black box, which offers a response when exposed to air, and when exposed top air containing odorants.

   The main objective of WG41 became the definition of validation procedures for IOMS instruments in the application. How predictive are the indicator values of an IOMS for the sensory perception of a human assessor?

   WG41 created a number of smaller Task Groups from its expert members, smaller groups with a very high focus which could identify areas of common understanding and areas where gaps in understanding and agreement existed, in an agile and efficient way. The results of discussions in the Task Groups was then reported back to the plenary meeting of WG41 for discussion.

   The following Task Groups were established:

  • TG1: Minimum requirements for instrumental odour monitoring
  • TG2: Validating the relationship odour metric and odour
  • TG3: Terms and definitions
  • TG4: Descriptions and review of scope relevant technologies

   In some areas, very good progress was made. Terms and Definitions, for example, have been drafted by TG3 and the objective was achieved to clearly separate between concepts of the sensory process of olfaction and the instrumental process of an IOMS determining “odour stimulus indicators”. Also, TG4 had a relatively straightforward task of creating a listing of existing technologies used in IOMS devices.

   However, the tasks of TG1 and TG2 proved to be more challenging. These two groups were joined in the later stages of development and the number of participating experts grew steadily, while not reaching conclusions on all outstanding tasks.

   Ample consideration was given to the context of the validation protocol. Most air quality monitors are calibrated using reference gases in a laboratory setting. However, for the broad concept of “odour” as a measurand reference materials are impossible to obtain. There are about 10.000 odorant compounds known to cause an olfactory response in humans. The odour detection threshold concentrations vary from a few ppt to ppm, 6 orders of magnitude. Considering this complexity WG41 reached the conclusion that the validation of IOMS had to be performed in the specific context of application, using the actual “odour under study” of a specific source or collection of sources, instead of a simplified reference material.

   Of course, non-odorant interferent quantities in the background matrix can also elicit a response from the IOMS. Not only the known influence quantities of humidity and temperature, but also potentially odourless constituents of ambient air, such as N2O or methane from agricultural operations, CO2 and CO from combustion engines in traffic etc. can act as odour metric indicator influence quantities. Considering these issues, the conclusion was that the validation needs to be performed in the context of application of the IOMS, for a specific source and source context. This approach has similarities to the approach in WG42, which is tasked with designing a validation method for air quality sensors. There, the sensors are validated paired measurement and subsequent comparison of results with a standardised reference method. The difference with odour is that a reference method for continuous measurement for odour does not exist, as the EN13725 method is suitable for discreet grab samples only.

   The validation for the measurement tasks “detecting absence or presence of odour” and “Odour classification” were tackled first, leading to draft clauses for the standard. However, the validation of the measurement task of measuring an odour stimulus indicator using an IOMS which could serve as an olfactory stimulus indicator value for odour concentration proved to be a more complicated challenge.

   Some WG41 members insisted that the standard had to contain a method to assess whether an IOMS was working according to its rated operating condition, as specified by the manufacturer.. A decision was made to use methods developed for laboratory testing in WG42 to design such a preliminary laboratory check. The outcome of this instrumental test would be: the instrument is functioning correctly, according to its rated operating condition, and can be used for the further validation test. Or the contrary, in case of a fail.

   For the validation testing protocol in the field, samples of (the source of) the odour under study are to be presented to the IOMS in known dilutions, using ambient air as the dilution gas. The odour concentration of this gas is measured by dynamic olfactometry according to EN13725 as a reference method. The results are then compared.

   The difficulty is that the validation test programme has practical limitations, in time and in cost. This means that it is difficult to collect a sufficient dataset to accommodate the significant inherent variance in both the reference method, olfactometry, and the IOMS method that is being validated. An additional challenge is the variations which may occur in the gas composition of the source, which may not affect the odour concentration, but which can cause additional variance in the IOMS results.

   This topic is as yet unresolved. The broad picture of the statistical design of the test method has been achieved. Because the actual frequency distribution of the measurement results of the IOMS are not known, the statistical treatment of the results must use a non-parametric approach. For this reason, the Chebyshev estimation of variance has been identified as the best method to derive a variance estimate from the validation results. The downside of this approach is that it is very conservative, risking an overestimation of the variance. This property of the statistical approach combined with the inherent large variance in both the olfactometric reference values and the IOMS results to be validated combine to risk making the test meaningless, because the degree of agreement is underestimated. In the final stages of the WG41 meetings and the work of the combined TG1/2 Monte Carlo simulation techniques have been used to explore the optimum number of samples and replicate analyses to achieve a meaningful result, given a certain claim by the manufacturer.

   A likely outcome is that the size of the dataset will need to be determined based on the manufacturers claim. For example, the requirement could be that the variance in the reference data would need to be a factor smaller (half?) than the variance in the manufacturers claim.

   This fundamental issue needs further deliberation and resolution before we can safely plan the path to completion of the standard.

 

3. Conclusions

   In conclusion, we can observe that:

  1. Working group CEN/TC264/WG41 has not been able to achieve its objective of a standard for validation of continuous measurement of odour using IOMS within the 6-year deadline

  2. Significant progress has been made in drawing up the conceptual framework, the terms and definitions and methods to validate the measurement tasks for determining the absence/presence of odours and the classification of odours under study

  3. While the statistical approach and the validation protocol in the field in the context of application has been largely defined, the optimisation of the test protocol for validating the olfactory stimulus indicator value of an IOMS as compared to odour concentration remains an unresolved challenge.

  4. Once the Work Item is formally reactivated, the optimisation topic needs to have been progressed and an outline of a proof of concept trial for application and validation of the method in a practical setting needs to be defined.

 

4. References

   Bell, A.G. 1914, Discovery and invention by Alexander Graham Bell, Reprinted from the National Geographic Magazine, June, 1914 Washington, D.C. Press of Judd & Detweiler, Inc.1914

   Gardner, J.W.; Bartlett, P.N., 1994. A brief history of electronic noses. Sens. Actuat. B: Chem. 1994, 18, 211-220

   Schulz, T.J., van Harreveld, A.P., 1996. International moves towards standardisation of odour measurement using olfactometry, Water Science and Technology, Volume 34, Issues 3–4, 1996, Pages 541-547

   Zwaardemaker, H., (1895), Die Physiologie des Geruchs, Verlag from Wilhelm Engelmann, Leipzig, 1895.

All the content here under Creative Commons license