Applying the electronic nose in the environment : requirements for the instrument
33
0
0
Texte intégral
(2) Research group "Environmental Monitoring" Electronic noses. Recognising Monitoring gaseous ambiences in the environment GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(3) Recognising gas ambiences Signal patterns from an array of sensors fresh waste. engine exhaust gas. odour neutralizer. landfill gas. compost. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(4) Monitoring gas ambiences. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(5) Array of tin oxide sensors. Drawback : heated above 300°C Advantages : robust, commercially available, “non selective” GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(6) First kind of application : pure gases or simple gas mixtures. E T H A N O L. CO H2 S. Interest . Testing the performance of the array Testing different mathematical procedures Testing different operational conditions Alternative when the sensor doesn’t exist or is expensive (BTEXÎ Benzene, Toluene, Ethylbenzene, Xylene). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(7) Second kind of application : head space above liquids, food, …. Problem even more difficult : •variable gas composition. But :. •same main chemicals •head space, reproducible GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(8) Third kind of application :. real atmospheres sampled in the environment. Problem of field sampling : •ever-changing chemical mixture •risk, uncertainties •influence of ambient parameters. But :. •Reproducible laboratory conditions. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(9) Fourth kind of application :. measurement of real atmosphere directly in the field. All the difficulties are cumulated : •ever-changing chemical mixture •risk, uncertainties. •influence of ambient parameters •Non-reproducible operating conditions GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(10) Additional difficulty : Odour assessment Research group « Environmental Monitoring » : Applying the electronic nose principle (with tin oxide sensors) to recognise and to monitor real life malodours, in particular in the environement, and directly in the field. Aims : •Understanding the odour release •Controlling odour abatement techniques. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(11) Obstacles to the monitoring of real life environmental odours. •The final goal of the study •The analysed sample •The operating conditions. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(12) Obstacles to the monitoring of real life environmental odours. •The final goal of the study Evaluation of a global odour annoyance (not concentrations in chemicals). •The analysed sample •The operating conditions. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(13) Measuring an odour :. 1. Suitable choice of the sensors 1. 1. 1. H2 S. R/R0. NH3. R/R0 0.1. 0.1. R/R0. TGS 824. TGS 825. TGS 2620. 0.1. CH3CH2OH. 0.01. 0.01 1. 10. 100. Concentration H2S (ppm). 10. 100. 1000. Concentration Ammonia (ppm). 0.01. 10. 100. 1000. Concentration Ethanol (ppm) 1. 1. R/R0 0.1. C3H8. R/R0. TGS 813. 0.1. TGS 822 0.01 100. 0.01. 1000. Concentration Propane (ppm). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. 10. 100. 1000. Concentration Benzene (ppm) Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(14) Measuring an odour :. 2. Supervised pattern recognition with "odours" as targets. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(15) Five typical malodorous sources Gas sampled in Tedlar® bags, near the source (emission level), tests in the lab (12 Figaro® sensors) Rendering plant. Waste water treatment plant. 59 samples – 7 months (MarchOctober) – various climate conditions – various operating conditions. Paint shop in a coachbuilding. Urban waste composting. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Printing house. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(16) Influence of the sampling time 8. 4. September. April. 6. March. Root 2. 2 0 -2 -4. June. October. -6 -8 -15. July -10. -5. 0. 5. 10. Root 1 GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium). 15.
(17) Influence of the odour source TGS824 ammonia TGS825 H2S TGS800 food vapours. TGS822 solvents TGS800 cigarette TGS813 combustible gas. TGS824 ammonia TGS2180 water vapour. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(18) Linking the sensor signals to the odour intensity (operator "feeling"). • near the fresh waste or near a biogas extraction well • 141 observations (69 « fresh waste » and 72 « biogas ») • feeling of the odour intensity on a Partial Least Squares regression (PLS) 4 level scale 71% of good predictions GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(19) Linking the sensor signals to the odour intensity (olfactometry). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(20) Emergence of an "odour event" 20. 10. Cl assi f i cat i on f u n ct i on s val u es. "odour event" pointed out 15. odor event. 10. 5. 0. -5. 8. The classification function for biogas decreases. 6 4 2 0 -2 -4. The classification function for fresh waste remains nearly stable. -6 -8. -10. 0:00. 2:00. 4:00. 6:00. 8:00. 10:00. 12:00. 14:00. 16:00. 18:00. time (hours). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. 20:00. 22:00. -10. M oving away from the landfill gas extraction well. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(21) Obstacles to the monitoring of real life environmental odours. •The final goal of the study •The analysed sample Highly variable (process, influence of environmental parameters). •The operating conditions. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(22) Group overlapping : various "purity" levels. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(23) Group overlapping : sensor drift. 1998 Print house. Waste water (blue). (mauve). Compost. 1999 2001. (red). Paint shop (green). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Correct classification : •1998 : 97.9 % •1999 : 81.8 % •2001 : 20.0 %. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(24) Sensor calibration ?. Sensor resistance. • Which standard gas ? • Ethanol = single gas • What is the « concentration » of the odorous gas mixture ? • Our solution : equivalence between the « concentration » of the odorous gas mixture and the concentration of ethanol. Odorous gas. Ethanol. Ethanol concentration GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Odorous gas dilution Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(25) Sensor "calibration". Assessing the « concentration » of odorous gas and the detection threshold of the sensors. E T H A N O L. Ethanol. Odorous gas (compost). Î 1 … 25 ppmv ethanol-equivalent. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(26) Low concentration level. 10 ppm … 1 ppm … 100 ppb … 10 ppb. Improving the sample uptake : e.g. pre-concentrating the analytes prior to investigation with the e-nose by a « field pre-concentrator » Nieß et al. Sensors & Actuators B (2003) 95 1-5. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(27) Obstacles to the monitoring of real life environmental odours. •The final goal of the study •The analysed sample •The operating conditions Far from any building, not easy to reach. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(28) Field instrument • • •. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Simple Transportable Reduced maintenance. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(29) Reference to the base line. sensor resistance. R0 = “base line” obtained with “pure” air. Best classification in our case. signal. not (R0-R) but R. time R = “signal” obtained with odour sample. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(30) Non-availability of the electrical supply network. 12 « Figaro » sensors. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. • 10 watts with « TGS800 » series • About 3 watts with « TGS2000 » series • with thin film sensors Î a few hundreds mWatts Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(31) Influence of ambient parameters •Air humidity •Temperature •Wind speed The air humidity takes part of the global odour : not possible to dry or to saturate the sampled gas. Water vapour. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(32) Test on synthetic odours mixtures of alcohols, esters, amines, aldehydes, ketones, sulfides prepared in Tedlar bags under uncontrolled external conditions (various humidity level) Training set of signals Good generated by recognition of odour at 6 new samples any humidity level Training set of signals generated by odour at low humidity level. Neural network (18 log-sigmoïd neurons, backpropagation). GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Unable to recognise 6 new samples at higher humidity. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(33) Conclusion •. electronic nose with very simple configuration leads to promising results. •. hope of designing a portable instrument to predict an unknown odour, “on line” in the environment. •. monitoring of environment = challenge, but a rough estimation is sufficient Î no need for restricting operating conditions. •. further work is still needed. •. Î for the sensors : improvement of the sensibility, the reproducibility, the electrical consumption, the drift. GOSPEL Artificial Olfaction Showcase Tübingen, June 13 – 14th 2006 Applying the electronic nose in the environment. Jacques NICOLAS – Anne-Claude ROMAIN Julien DELVA University of Liège (Belgium).
(34)
Documents relatifs
As a result of the study on the possibilities of applying parallel accounting for long-term natural resources, we propose to account the technogenic gaseous
The issue of singular- ities at low frequencies remains, but can be treated by either adding assumptions on the interaction potential w, as in [9] (allowing a large class of
Il établi une liaison covalente et deux liaisons hydrogène avec deux atomes accepteurs qui sont liés au même atome X [25, 26] (figure 4a), la figure 4b présente aussi
modélisation fonctionnelle qui permet de relativiser l’intérêt de certaines prouesses algébriques. Le principe curriculaire : avancer plutôt que de reporter, dans le
Biogas Production Potential of Co-digested Food Waste and Water Hyacinth common to the Niger Delta..
AMBER is a near infrared, three beams, dispersed fringes, single mode VLTI focal instrument.. Besides a very wide scientific program, the near infrared, and particularly the K
In this thesis, a digital controller for a Unity Power Factor AC-DC converter is designed, based on a linear large-signal model of the power supply.. A
In order to satisfy criterion (4), in NIPALS, PLS-R and new Mode A PLS-PM, when work- ing on standardized variables, optimal weights are calculated as Pearson’s
