CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, Wind Tunnel Simulation of Air Pollution Dispersion in a Street Canyon

Transcription

1 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, SPECIAL GUEST EDITOR SECTION Wind Tunnel Simulation of Air Pollution Dispersion in a Street Canyon SVATOPLUK CIVI and MICHAL ST I K J. Heyrovský Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Dolej kova 3, Prague 8, Czech Republic ZBYN K JA OUR and JAN HOLPUCH Institute of Thermomechanics, Academy of Sciences of the Czech Republic, Dolej kova 5, Prague 8, Czech Republic ZDEN K ZELINGER J. Heyrovsk Institute of Physical Chemistry, Academy of Sciences of the Czech Republic, Dolej kova 3, Prague 8, Czech Republic Physical simulation was used to study pollution dispersion in a street canyon. The street canyon model was designed to study the effect of measuring flow and concentration fields. A method of CO 2 -laser photoacoustic spectrometry was applied for detection of trace concentration of gas pollution. The advantage of this method is its high sensitivity and broad dynamic range, permitting monitoring of concentrations from trace to saturation values. Application of this method enabled us to propose a simple model based on line permeation pollutant source, developed on the principle of concentration standards, to ensure high precision and homogeneity of the concentration flow. Spatial measurement of the concentration distribution inside the street canyon was performed on the model with reference velocity of 1.5 m/s. At present, a number of experimental techniques are based on the principle of in situ measurement of atmospherically interesting molecules in the gas phase. Most of these techniques are technically and financially exacting. For example, MIPAS (1) or ALIAS-II (2) experiments belong to the category of instruments flown under lift balloons with a remote sensing device, applying emission Fourier transform or laser absorption spectroscopy techniques in the IR spectral range. At present, similar experiments also use satellites (3), which yield a high spatial resolution and a steady monitoring process (lasting several years) of the stratospheric molecules over the earth. The ground observations were extended from earth s atmosphere into terrestrial or interstellar space. A very sensitive microwave detection technique (due to atmospheric window) together with large telescopes provide Guest edited as a special report on Environmental Analysis by Joseph Sherma. Corresponding author s civis@jh-inst.cas.cz. high sensitivity and image resolution ability, adding much to our knowledge of space chemistry and the composition of the universe (4). The methods of laboratory modeling are in contrast to such very complicated in situ detection techniques. These methods, which are based on the principle of physical and mathematical modeling (5), use small models in an aerodynamic tunnel. They provide useful information about the dispersion of selected pollutants in urban agglomerates, which are frequently difficult to determine by direct monitoring methods in the lower part of the troposphere, designated as the atmospheric boundary layer (ABL). Simulation of pollutant dispersion on the model of future construction in a given agglomeration can prevent urban mistakes, which could cause local accumulation of toxic gases in urban air and thus directly endanger future inhabitants. Qualitative studies of a model of atmospheric processes in wind tunnels can be performed by using various methods to visualize the flow and dispersion of molecules replacing the pollutant (5). For real estimation of the concentrations, a flame-ionization method is often used. High sensitivity methods are important for extending spatial resolution, and for obtaining quantitative information about concentration profiles in the studied areas. The method must be dynamic in a wide range of concentrations and must be selected for the kind of molecule studied. We applied CO 2 absorption laser photoacoustic spectrometry as an analytical tool to monitor gas dispersion in a wind tunnel (6). The advantage of this method is its high sensitivity and broad dynamic range, permitting monitoring of concentration from trace values to saturation. For simulation of the atmospheric processes in a wind tunnel, we chose the methanol molecule as a model gas. Methanol has a strong absorption vibration-rotation band in a spectral region of 10 m. This absorption coincides with the emission frequency of a CO 2 -laser, which can be used for stimulation of a photoacoustic effect. The main advantage of the 10 m spectral range is in the presence of an atmospheric window where absorption of atmospheric water and CO 2 in air is negligible.

2 244 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, 2002 Figure 1. Experimental setup for concentration measurements of air pollution in the wind tunnel. Laser Photoacoustic Detection Classical absorption spectroscopy measures the intensity between the in-coming and out-going radiation. Thus, in determination of low concentrations of the absorbing molecules, very small differences between the intensities of I 0 entering and I leaving radiation are detected. In the case of a laser where the radiation intensity is large, on a relatively large offset signal, relatively small intensity changes are detected: I 0 I=N LI 0 (1) where (I 0 I)<<I 0,I 0 > I, N is the number of absorbing molecules, is absorption cross section, and L is the length of absorption space. It is much more efficient to detect a small signal without the presence of the large offset background. One way to solve this problem is to detect the absorbed energy by means of the laser photoacoustic method. The laser photoacoustic detection method uses the so-called photoacoustic effect, first described in 1880 by Alexander Graham Bell (7). When electromagnetic radiation passes through an absorbing environment, photon absorption occurs. An excited molecule can lose energy principally in 3 ways: in the radiation process, in transfer of the vibration energy to another molecule, or in collision with a molecule when the energy of absorbed radiation is transferred to kinetic energy of both participating particles. Under atmospheric pressure, mainly the last of the specified mechanisms applies, i.e., the so-called vibration-translation relaxation. This leads to a local increase of gas temperature in place of absorption. If the radiation is chopped, then pressure strokes occur in the photoacoustic cell with the frequency f 0 equal to the frequency of the chopping. A value of the modulation frequency f 0 has to be suitably selected, taking into consideration the relaxation time needed for collision redistribution of the energy absorbed by the molecules. Then, occurring acoustic modulation can be detected with a sensitive microphone. Thus, we can formally divide the photoacoustic effect into 2 phases: absorption of the electromagnetic radiation by the molecule, and the following generation of acoustic waves in the cell. The generation of acoustic waves as a function of the energy released by the excited molecule is described by a nonhomogeneous wave equation (8). If the condition of the sufficiently small intensity of exciting radiation is held (the absorption transition is not saturated) and the modulation frequency f 0 is chosen so that f 0 1 ( represents the time necessary for collision redistribution of the energy absorbed by the molecules), detected photoacoustic signal S can be expressed as (9): S = CPN where P corresponds to the power of laser radiation, N is a number of absorbing molecules per cm 3, is an absorption cross-section of an absorbing molecule, and C ( 1)GLQ1 R mic p 1 (r mic ) fv 0 0 is a constant characterizing the properties of the cylindrical photoacoustic cell. In equation 3, is Poisson constant, G is a geometrical factor, L is the length, V 0 the volume of the cell, Q 1 is the quality factor of the first acoustic mode p 1, f 0 is a modulation frequency, and R mic is the sensitivity of the microphone placed at distance r mic. From equation 2 it is obvious (3)

3 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, Figure 2. that the sensitivity of laser photoacoustic detection increases with the increasing power of laser radiation. The photoacoustic effect has been known for more than a century; the first gas analysis using photoacoustic detection was, however, described 50 years later. In 1943, the detection limits of analyzed gases were successfully reduced from several per mille to ppmv, and this limit was not exceeded for a number of years because of the low power of radiation sources and the low sensitivity of detectors then used. A renaissance of photoacoustic detection came with lasers. In 1968, Kerr and Atwood were the first to describe laser photoacoustic detection. They used continuous CO 2 -laser to determine trace concentrations of CO 2 in the inert nitrogen atmosphere (10). In 1971, Kreuzer determined the detection limit of methane by using He-Ne laser emitting at 3.39 m (11). With his collaborators from the Hewlett-Packard Laboratories, he also developed a CO/CO 2 laser photoacoustic analyzer and determined detection limits of many other gaseous substances (12, 13), and dealt with the use of the photoacoustic detector in gas chromatography (14). An innovation leading to the efficiency of detection was the development of the acoustic resonant cell (15). Experimental Scheme of the photoacoustic cell used. The experimental set-up used is schematically depicted in Figure 1. A model of a symmetric street canyon perpendicularly orientated to the wind direction was inserted into the aerodynamic tunnel (cross-section, cm), which provides a simulation of processes running in the atmospheric boundary layer. The street canyon model used was 70 cm wide, with wall heights of 70 cm. At the bottom of the canyon, a line source of pollution was located to simulate cars in a street. As a line pollution source, we used a set of silicone tubes, sealed on both ends with a steel ball and filled with 99.9% methanol (Lachema, Neratovice, Czech Republic). Permeation of methanol through the walls of tubes generated gaseous contamination inside the canyon. The silicone tubes used (Deutch and Neumann, Germany) were 150 cm long, cm od, and with a wall 1.5 mm thick. The used permeation tubes enabled generation of methanol concentrations in the range of to g/s. Continuous monitoring of the atmosphere inside the canyon was ensured by the use of a photoacoustic gas analyzer. The source of radiation was a tunable 12 CO 2 -laser with internal modulation (Edinburgh Instruments, Edinburgh, Scotland, WL-8-GT), emitting on rotation-vibration transitions of the CO , in the spectral range between 9 and 11 m. Tuning of the 9P34 laser emission line, which was used to calibrate the equipment and for the analysis itself, was being checked during the whole time of measurement by the spectrum analyzer (Optical Engineering, Inc., Fremont, CA; 16-A Model). Optimum modulation frequency of the laser f 0 = 1.32 khz was chosen experimentally. With this frequency, the value of the detected photoacoustic signal was in acoustical resonance. The laser radiation, after its passage through the photoacoustic cell, was impinging on the active surface of the pyroelectric detector, and thus its intensity could be controlled. Samples of the atmosphere contaminated with methanol were taken from individual, precisely defined places of the canyon by means of a movable probe formed by a Teflon tube, bringing the analyzed gas at a speed of 2 3 cm 3 /s into the photoacoustic cell. A schematic draft of the cell used in all measurements is shown in Figure 2. The central part of the cell forms a cylindrical absorption space 8 mm id, in which interaction occurs between gas particles flowing through and laser radiation. A created photoacoustic signal is detected by a sensitive microphone inserted into the body of the cell at approximately half the absorption length. Buffer spaces located behind the input and in front of the output window eliminate parasite acoustic signals occurring as a consequence of the absorption of radiation in the windows of the cell. The cell used (380 mm long) is made of brass, and KBr windows (7 mm thick, with a diameter of 50 mm) were transparent for IR radiation of wavelength ca 17 m (loss by reflection and absorption represents 10% of radiation power). During the measurement, the cell was tempered to 30 1 C. To detect pressure changes inside the cell, the electret microphone (TPR 175 E, Tokyo, Japan) was used. The generated photoacoustic signal detected by microphone was further electronically processed by the digital synchronous amplifier and evaluated by computer. The permeation method (16) was used not only to generate trace concentrations of methanol inside the street canyon, but also to calibrate the photoacoustic analyzer. Concentration standards, long-term weighted polyethylene tubes filled with methanol and sealed on both ends with steel balls, located in the permeation chamber inserted in front of the photoacoustic

4 246 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, 2002 a b Figure 3. Resulting spatial distribution of model pollutant in absolute (ppmv) and dimensionless concentration scale for (a) one-road model, (b) 2-lane motorway model. cell, allowed generation of gaseous methanol in trace concentration c (in ppmv): c 10 6 VM vperm (4) F M where V M is the molar volume of the carrier gas flowing through the permeation chamber (cm 3 /mol), perm (in g/s) represents the mass loss of the standard s filling per time unit, M (in g/mol) is the molar mass of the standard s filling (for methanol g/mol), and F is the flow rate of carrier gas (in cm 3 /s). The permeation flux depends on the length, inner and outer diameter of the tube, the material the standard is made of, the partial pressure of the generated gas inside and outside the tube, and the temperature. Mass losses of the filling were measured long-term. Because of the temperature dependence of permeation, the calibration standards, in the period between weighing and measuring, were kept in a room with a constant temperature of C. The generated concentration from calibration standards was corrected to the actual temperature in the wind tunnel. The expiration time of standards prepared was determined by the time interval in which it was possible to regard the permeation speed as constant. For the calibration standards, it was tens of days; the expiration time of silicone tubes was substantially shorter, in tens of hours. Results and Discussion The photoacoustic analyzer was calibrated by measuring the dependence of detected photoacoustic signal on the concentration of methanol generated in trace amounts by permeation standards (equation 4). The detection limit of methanol c L = 0.07 ppmv (96 g/m 3 at T = 292 K) was for emission line 9P34 of 12 CO 2 -laser (wavelength = m) calculated on the base of the relation (equation 5): c L sk = q s 2 2 a / k a +s B +( ) 2 2 (5) k where k is the slope of the regression straight line with the standard deviation of determination, s k, s a is the standard deviation of the particular intercept a on the axis of detected signal, s B represents the standard deviation of the measured back-

5 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, ground signal, and q = 3 is a numerical factor chosen according to the required level of statistical significance (1 ) = 0.99 (17). Our obtained values correspond with methanol concentration levels in air contaminated by industrial pollution sources. An ambient methanol concentration of about 140 g/m 3 (18) is detectable at the present adjustment of our apparatus. Higher values, from 280 to g/m 3 can be caused by large industrial sources, e.g., pharmaceutical plants (19). That means use of our analyzer both for in situ monitoring of methanol concentration in burden regions and for on-line air control at particularly exposed workrooms (~ ppmv) is possible because of the broad dynamical range. Our objective was the study of spatial distribution of pollutants coming from intensive vehicle traffic simulated inside the street canyon. This study is closely connected with our previous measurements (6) aimed at the quantitative description of pollutant distribution at the bottom and at street canyon walls, depending on wind speed. The main emphasis of this work was to investigate spatial pollution dispersion in the whole volume of the street canyon. Models simulating a simple road and a double lane motorway were investigated. Parameters of used models were transverse length L = 150 cm, height H = 70 cm, and distance between walls 70 cm. All experiments were performed at wind speed U = 150 cm/s. For the model with one lane, a permeation source of pollutant was located symmetrically in the middle of the street canyon. For the double lane motorway model, the sources were placed in one-third and two-thirds of the street canyon s bottom. Temperature in the aerodynamic tunnel corresponded to the room temperature and varied from 18 to 25 C. Therefore, we used several sets of permeation sources calibrated to actual temperature. To compare our measured data with calculated models, the absolute concentration values (depending on flux Q(T) of permeation source just used) were converted into the dimensionless concentrations X by the relation: X KHUL Q where K is the concentration (in g/cm 3 ), Q permeation flux (in g/s), H (in cm) and L (in cm) are geometrical characteristics of used model defined above, and U = 150 cm/s is the actual wind speed in tunnel. (a) Model of a simple one-lane road. Methanol concentration levels were measured in the whole volume of the street canyon. Horizontal distances between individual sampling points were 5 cm (at 0, 5, 7, 10, and 20 cm from the bottom); near the top of canyon they varied from 5 to 20 cm (at the height of 30, 40, and 50 cm). Figure 3 depicts the range of detected concentrations from 1.5 to 100 ppmv for the simple road model. Figure 3a depicts the gained concentration profile, showing that the highest concentrations at the bottom of the street canyon near the permeation source reached values of ppmv. The increased concentration on the leeward side of the street is also obvious. By contrast, the opposite windward side is ventilated with greater efficiency. This fact relates to the formation of circulation currents, which move (6) the contamination into the leeward side of the street where it accumulates. This is in accordance with the results of the mathematical simulation of considered processes (5). Mathematical models predict that the concentration decreases with increasing wind speed and, furthermore, concentrations at the leeward side of the street are higher than those at the windward side. These are the most essential features of pollutants dispersion in street canyons, and therefore the street model, with some minor modifications, is still widely used (especially for engineering applications). Any more detailed description cannot be made on the basis of a simple mathematical modeling. An essential drawback of the model is its very rough parameterization of wind direction dependence. Moreover, with decreasing wind speed, uniform concentration distribution across the street canyon may be assumed. Mathematical models generally do not describe these properties. Therefore they are not recommended for wind speeds lower than 1 m/s. Under these conditions, experimental modeling plays the dominant role. (b) Model of a double-lane motorway. Figure 3b shows the same effect of pollutant accumulation in principle. Near the bottom of the street canyon where contamination sources are located (one-third and two-thirds of the street canyon s bottom), the situation is similar to that of the single-lane road model. At permeation sources, the methanol concentration is the highest, reaching values exceeding 100 ppmv, with a permeation rate of the used sources of g/s ( 10%). The figure clearly suggests the great magnitude of contamination to which drivers of slowly moving cars in both lanes of the motorway are exposed as well as the people walking at the leeward side of the street canyon. Toward the leeward side, the contamination again increases, filling the bottom of canyon and reaching, in an order of magnitude, a smaller concentration only in half of the street canyon. The windward side is better ventilated. The effectiveness of ventilation increases with each increase of the distance of the line source, however small. Conclusions The main goal of this work is an application of laser photoacoustic detection for simulation of the dispersion of atmospheric pollution. The simulation performed in a wind tunnel using a simple model of a street canyon with a line permeation pollution source. The line permeation source simulated the emission of the pollution caused by automobile traffic. A model of simple (one-lane) street and (double-lane) motorway was used in a number of measurements of spatial profiles across the street canyon to demonstrate the differences in ventilation of air pollution from road traffic. The obtained results are in very good agreement with mathematical numerical models based on the solution of diffusion equations. The advantage of the applied method of laser photoacoustic detection consists in its high sensitivity and broad dynamic range, permitting monitoring of concentrations from trace amounts to saturation values. The use of this method permitted proposal of a simple model of a linear permeation pollutant source, functioning on the principle of concentration standards, to en-

6 248 CIVI ET AL.: JOURNAL OF AOAC INTERNATIONAL VOL. 85, NO. 1, 2002 sure high precision and homogeneity of the concentration flow on a permeation basis. Spatial measurements were performed on the street model of the concentration field inside the street canyon for a reference velocity of 1.5 m/s. The measurement demonstrated that the maximum concentration is located at the bottom of the street in the vicinity of the source toward the leeward wall. Acknowledgments This work was funded from the budget of the Czech Republic by means of the Ministry of Education, Youth and Physical Training of CR (in the frame of COST 715) and of the Grant Agency of the AS CR under grant No. A References (1) Oelhaf, H., Fischer, H., Wetzel, G., Stowasser, M., Friedl-Valon, F., Maucher, Q., Trieschmann, O., Ruhnke, R., & Sasano, Y. (1998) Proc. SPIE 3501, (2) Scott, D.C., Herman, R.L., Webster, C.R., May, R.D., Flesch, G.J., & Moyer, E.J. (1999) Appl. Opt. 38, (3) Sasano, Y., Suzuki, M., Yokota, T., & Kanzawa, H. (1995) SPIE J. 2583, (4) Smith, D. (1988) Philos. Trans. R. Soc. London. A324, (5) Berkowitz, R. (1998) in Urban Air Pollution-European Aspects, J. Fenger, O. Hertel, & F. Palmgren (Eds), Kluwer Academic Publishers, London, UK, pp (6) Zelinger, Z., Civiš, S., & Ja our, Z. (1999) Analyst 124, (7) Bell, A.G. (1880) Am. J. Sci. 20, (8) Kreuzer, L.B. (1977) in Optoacoustic Spectroscopy and Detection, Y. Pao (Ed.), Academic Press, New York, NY, pp 1 25 (9) Sigrist, M.W. (1994) in Air Monitoring by Spectroscopic Techniques, M.W. Sigrist (Ed.), John Wiley & Sons, New York, NY, pp (10) Kerr, E.L., & Atwood, J.G. (1968) Appl. Opt. 7, (11) Kreuzer, L.B. (1971) J. Appl. Phys. 42, (12) Kreuzer, L.B., Kenyon, N.P., & Patel, C.K.N. (1972) Science 177, (13) Kreuzer, L.B. (1974) Anal. Chem. 46, 235A 244A (14) Kreuzer, L.B. (1978) Anal. Chem. 50, 597A 606A (15) Dewey, C.F., Kamm, R.D., & Hackett, C.E. (1973) Appl. Phys. Lett. 23, (16) Barrat, R.S. (1981) Analyst 106, (17) Long, G.L., & Winefordner, J.D. (1983) Anal. Chem. 55, 712A 724A (18) Meyer, P.L., & Sigrist, M.W. (1990) Rev. Sci. Instrum. 61, (19) Sigrist, M.W. (1994) Analyst 119,