You are on page 1of 24
a function of distance downward. erations, One particular solution to this equation is known as the Fickian diffusion equation, which predicts the pollutant concentration to be binormally distributed. ion of the Fickian diffusion equation applied to the atmospheric dispersion problem has been approximate physical reality. Another valid approach is based on th of the dispersion pro- cess. ‘This model is usually referred to as the Gaussian dispersion Sduation and is discussed in the next section > 20.3 The Gaussian Model ch behavior hae bes oles about a plume conter- Ie Gouean been shown by Pasquill (1961) to Perel modeled by a double Gaussian equation, Then equation (which ale the Ssein ot in aeactive gaseous pollutant from 4 conte giauree) is given here in a form that predicts th - Sentration ata point ,y, 2) located downwind from the cee on 1 a Pn NNR exp] — ( 656 Chapter Twenty Atmospheric Dispersion Modeling where C = steady-state concentration at a point (x,y,z), we/m® Q = emissions rate, jlg/s 0,, 6, = horizontal and vertical spread parameters, m (these are functions of distance, x, and atmospheric stability) u = average wind speed at stack height, m/s ¥ = horizontal distance from plume centerline, m 2= vertical distance from ground level, m H = effective stack height (H = h + Ah, where h = physical stack height and Ah = plume rise, m) The double Gaussian distribution in the plume modeled by Eq. (20.1) is shown schematically in Figure 20.4, Note that there are two terms in Eq, (20.1) with an exponential in 2, one with (z - H) and one with (z + H). The reason for the two terms is to account for the fact that pollutants cannot disperse under- ground. A further explanation for this apparently trivial statement is in order, N 2, one with (z — H) and one with (z + H). The reason for the two terms is to account for the fact that pollutants cannot disperse under- ground. A further explanation for this apparently trivial statement is in order bu -y.2) bx, 0, 0) (-¥, 0) Figure 20.4 Coordinate system s! and vertical (Adapted fom Turner, 1970.) wing Gaussian distributions in the horizontal Section 20.3 ‘The Gaussian Model 657 Without the second exponential term in z, Bg. (20.1) would be mu (20.2a) e which is a “true” double Gaussian equation, with a source at y = 0, and z= H, The situation represented by Eq, (20.2a) is depicted in Figure 20.5(a), ‘which shows pollutant dispersing infinitely in the #2 direction including underground. Since this is impossible, the model is corrected by adding « ground by the real source, The equation for the image source is When Eqs, (20.28) and (20.2b) are added, Eq. (20.1) results. This behavior is depicted in Figure 20.5(b). As a convenience, we usually just say that the plume reflects at the ground level or exhibits reflection. If the plume were dispersing over water and contained a highly soluble gas, the assumption of reflection might not be justified. However, under usual conditions, it is assumed that pollutanis such as CO, VOCs, SO, or NO, do reflect at the grou Tt is important to keep in mind some general relationships indi- ceated by Bg, (20.1): 1. The downwind concentration at any location is directly propor- tional to the source strength, @. developed by the U.S, Environmental Protection Agency for atmo- spheric dispersion modeling. For this reason, a brief discussion of atmospheric stability estimation and some illustrations of the use of the Gaussian equation are presented in the following few paragraphs. Region of mathematical dispersion underground , Region of reflection iH Ny Section 20.3. The Gaussian Model 659 Concentration Downwind distance Figure 20.6 Variation of ground-level centerline concentrations with distance down- wind from an elevated source. Atmospheric Stability Classes ‘As was explained in Chapter 19, air is termed unstable when thore is good vertical mixing, This occurs with strong insolation (solar radiation striking the earth’s surface) combined with light winds. As a tesult of absorbing solar energy, the earth’s surface heats and then warms the layers of air near the ground. The warm air rises, promot- ing vertical mixing. Stable air results when the surface of the earth is cooler than the air above it (such as on a clear, cool night). ‘Then the layers of air next to the earth are cooled, and no vertical mixing can occur, Because the dispersion parameters, oy and oz, are str0ng fune- Hons of atmospheric stability, as well as downwind distance, it is 0.1 1 10 100 Distance downwind, km Figure 20.7 and partially neutral (U, D, and 1s) elasses, ‘therefore, tor the more com- 5000 Figure 20.8 662 Chapter Twenty Atmospheric Dispersion Modeling Table 20.1 Stability Classifications” Day Night CETL Solar Radiation __Cloudinoss® Wind Speed Cloudy Clear mis Strong? Moderate? slight (24/8) (38) 2 a 8 E F 23 AB c E F 25 8 © ° E ee c D D o > e ° D D 7 Saface wind spood is measured at 10 m above tho aroun ® Correspons to csr simmer day wh sun higher tna 60" above the hozon Corresponds to a summer day wha few trcken couse, ora oar day wil sun 25-60 ay wt tn 2 reas the fracon ashy covered by tous 8, 8-0, or c-D constons ‘mon neutral or near-neutral stability classes, utilizing the A wl would be more appropriate. Ultimately, most af iui's comparative results depended heavily ups th Moderately unstable E = Slightly stable C= Slightly unstable F = Stable Regardless of wind speed, Class D should be assumed for overcast conditions, day or night. Coo ee rs rN tt attimt eet ‘Adapted from Turner, 1970. mon neutral or near-neutral stability classes, utilizing the M-O length method would be more appropriate. Ultimately, most of Mohan and Siddiqui’s comparative results depended heavily upon the assumption that the P-G method gave accurate predictions, Small errors in measur- ing wind speed and/or slight differences among observers in judging the cloud cover or solar insolation, as well as site-specific variables relating to local geography and climate, can often result in variations of one sta~ bility class among different observers when using the P-G method. Often it is difficult for people to obtain consistent readings from Figures 20.7 and 20,8. Furthermore, such graphical representations are inconvenient for use in computer programs. Martin (1976) pub- lished equations that give reasonable fits to these curves, The general equations are (20.3) and (20.4) where a, b, c,d, and fare constants that are dependent on the stability class and on the distance x (x must be expressed in km), Section 20.3 The Gaussian Model 663 ‘The numerical values for the constants in Eqs. (20.3) and (20.4) are given in Table 20.2. The following example illustrates typical uses of the information presented in the past few pages. Table 20.2 Values of Curve-Fit Constants for Calculating Dispersion Cooficients as a Function of Downwind Distance and Atmosphoric Stability eee Tiicton of Downwind Distance and Atmospheric Stability Stability 2 be A 213 0.834 408 L841 9274507 B 156 0.804 414933 1082 © 104 0894 810 og 0 Bt. D 68 ogee 332 O725 17 aes E 505 08m 228 0678 13. 5 p 34 084 1435 0.740 035 G26 ‘Asapted rom Morin, 1976 rrr rrr Example 20.1 Nitric oxide (NO) is emitted at 110 gis from a stack ith physical height of 80 m, The wind speed at ‘80 m is 5 mis on an overcast morn: ing, Plume rise is 20 m. (a) Caleulate the ground-level centerline con centration 2.0 km downwind from the stack. (b) Caleulate the tration at 100 m off the centerline at th Solution Ty ETAT ‘Adapted from In, 1976. >rrrrrrr> Example 20.1 Nitric oxide (NO) is emitted at 110 ws from height of 80 m, The wind speed at 80 m a stack with physical ing. Plume rise 5 m/s on an overcast morn- 20 m, (a) Calculate the ground-level centerline con- centration 2.0 km downwind from the stac tration at 100 m off the centerline at the Solution As noted in Table 20.1, Class D stability should be used for o ‘cast conditions. From Eqs, (20,3) and (20.4), 0, = 126 m and o, Substituting all the proper values into Eq (b) Calculate the eoncen same x distance, 1) yields the following: (a) Aty = 0 and z= 0 (centerline and ground-ievel) (545)(0.21 159 jgim 664 Chapter Twenty Atmospheric Dispersion Modeling (b) At y = 100, everything else being the same: 2 C= 159exp| 2100" 1267 = 159(0.73) =116 jtg/m? 444440 __ In Chapter 19, we introduced the power law to predict the varia- tion of wind speed with height. The equation is repeated below for con venience: o 48 2) (19.1) my ly where 21) 2 = elevations 1 and 2 windspeeds at z, and 2 exponent The exponent p varies with atmospheric stability class and with surface roughness. For “rough” surfaces (typical of urban and subur- ban areas), the BPA (1995) recommends the exponent values shown in Table 20.3. For flat, open country and lakes and seas, there is less variation between the surface wind and the geostrophic wind. For use in Eq. (20.1) the wind speed should be adjusted to H, but to be conser- Ramnla 209 illns- vative EPA reeammende that 1 he adinated toh a———— a9.) where 2, 72 = elevations 1 and 2 uy, Ua = windspeeds at 2, and 2p = exponent ‘The exponent p varies with atmospheric stability class and with surface roughness. For “rough” surfaces (typieal of urban and subur ban areas), the BPA (1995) recommends the exponent values shown in ‘Table 20.8. For flat, open country and lakes and seas, there is less variation between the surface wind and the geostrephie wind, For use in Eq, (20.1) the wind speed should be adjusted to H/, but to be conser- vative, EPA recommends that u be adjusted to A. Example 20.2 illus trates the combined use of as. (19.1) and (20.1), Table 20.3 Exponents for Wind Profile (Power Law) Model Exponent (2) Stability Class Rough Surface (urban). Smooth Surface (rural) 045 045 0.20 025 030 030 amooo> ‘Rasped Wom US. Eneronmentl Protectan Agen, 1955 ror rrrrry Example 20.2 Consider the same data as in Example 20.1, except that the met orology is stich that: the wind speed (at 10 m) is 4 m afternoon on a hot summer day. Caleuk concentration at and it is mid fe the ground-level centerline 2.0 km, assuming rough terrain, Solution From Table 20.1, CI lass B stability is estimated, owing to the wind speed and strong insolation. To ealeulate the wind speed at stack height, we use Eq, (19.1). An estimate ofp for Class B stability is 0.15. Brom Bq, (19.1), we see that the wind speed at-80 m is (i) 5.5 mils From F es 20.7 and 20.8, estimates of, and c, are 300 m and 2210 m, respectively, Substituting into (20.1), we have roo)" 11008 c exp) 10 | 300)(230) 9°" |-3 92.3(91) C= 84 eho? we used qs. (20.8) and (20.4) to estimate the values of ‘we would have obtained 260 mu and 294 m, respectively. The final ansier then would have been 86 n/m Wee eeeeee From Figures 20.7 and 20.8, estimates of o, and 230 m, respectively, Substituting into Bq. (20.1), we have re 300 m and i ifpaxocao yt tl 1100? ] n(5.5)(300)(280) "| 2.230? | )2,3(.91), C= 84 je/m? Note that had we used Eqs. (20.8) and (20.4) to 9, and o, we would have obtained 290 m and final answer then would have been 86 pg/m?. estimate the values of m, respectively. The 14441 ‘The Dependence of Concentration on Averaging Time ‘The concentration predicted by ues from Figures 2 a. (20.1), using the o, and o, val 8 for Egs. (20.3) and (20.4)}, is a 10-minute-average concentration (although, to be conservative, the U.S. EPA treats these as 1-hour concentrations in its computer mod- cls). As mentioned pr would be expected fiously, a longer time-averaged concentration to be lower than a short time-average, owing to ‘wind shifts and turbulent diffusion. For averaging times between 10 min and 5 hr, data reported by Hino (1968) suggest that the concen- trations at two averaging times are related s follows: sGsiit (205) ESuMmaung We Maximum bownwind Ground-Level Concentration The maximum downwind ground-level concentration alway ‘occurs on the centerline, Turner (1970) developed a graphical means of estimating the maximum concentration (Cyax), and the distance at which it occurs (+ pa.)- Both Cjnax and x max depend on stability class, effective stack height, wind speed, and emission rate, Figure 20.9 presents Turner’s results as a plot of x max Versus (Cw/Qax: ‘The solid lines represent the various stability classes, and the numbers along the lines represent various values of the effective stack height. Typically, one knows the stability class, the effective height, the wind speed, and the emission rate. From these data and the plots, one may ¢ In order to better understand the qualitative variation of Cruoy and © max with stability class and effective stack height, look at Figure 20.10. Notice that as stability changes from unstable to stable, Cinax decreases slightly while x yay increases. Also, the region of relatively high concentrations is much larger under stable conditions. An increase in H results in a dramatie reduction in Craxs While inereasing X max Slightly. However, note that the concentrations a “long way” away from the source approach the same value regardless of stack height. ‘The plots in Figure 20.9 were fit to a polynomial equation by Ran. ehoux (1976), whieh ean be written as (C1 ®)pay = exp a+b (In Ht) + n +a(in 47] (20.6) where a,b, e and are constants that depend on stability lass as shown in Table 20.4. In Bq. (20.8), H must be in m and (Cu/Qmayy is in m=, Ono of the most straightforward methods to find Cry and tay i3 by the repeated use of Eq. 20.1), Of course, this method requires the use of a computor or programmable calculator, Typically, a very large x interval is chosen, inside of which the maximum is sure to aceur, The concentration is calculated at either end of the interval, using 668 Chapter Twenty Atmospheric Dispersion Modeling Unstable Neutral (a) (b) Figure 20.10 Behavior of ground-level centerline concentration from an elevated ———— (b) Figure 20.40 Behavior of ground-level centerline concentration from an elevated source as a function of downwind distance and (a) atmospheric stability (at constant H), and (b) effective stack height (at constant stability). Table 20.4 Values of Curve-Fit Constants for Estimating (CU/Q)may from fe Has a Funotion of Atmospheric Stability Constants b e 2718 2.1972 0.0389 ~113980 0 ~0.0934 o2tai method tor tnamg tne maximum concentration (ine method) has become easy to do. Simply start at x (20.1), then increment x by a small amount (say 100 meters) and solve the equation again, Repeat the process (in a spreadsheet this is almost absurdly easy) until the concentration is observed to pass through the maximum and starts to decline, brute roree 0 and solve Eq. rrr rrrry Example 20.3 Given the same data as Example 20.1, estimate the maximum ground-level concentration and the distance at whieh it oceurs. Solution For Class D stability and H = 100 m, from Figure 20.9, tnx is esti- mated to be 3 km, and (Cu/Q)mnax is estimated to be 8.1(10)-° m~2, Thus i 110 (10) Caz = 8.1(10)7 200 178 ug/m® Similar results should result from Eq, (20.6)—that is, o(10)° . O(10)"exp|-2.5802~1,9611n 100 ~0,0984(In 100)? 188 jug/m? Using the computer program listed in Appendix D, we obtained a value of 184 pgim? for Cas, which occurred at a distance of 2.9 km. We got identical results with the brute foree method using a spreadsheot. 444d Estimating the Downwind Concentration [TT 670 Chapter Twenty Atmospheric Dispersion Modeling Yel“ —e") He ay? anwoya,)"?( 2 of if iN ‘ \\ (20.7) ( eal \ exp er where L = height from the ground to the bottom of the inversion layer, m, In practice, it is only necessary to vary the summation index, j, from ~ 2 to + 2 to obtain reasonable convergenee of Eq. (20.7). During the break-up of a ground-based inversion, fumigation ofa plume can occur, Fumigation describes the situation in which an unstable boundary layer grows up to a fanning plume and quickly mixes it vertically throughout the distance between the ground and the plume. Under these conditions, t vation that best predicts the downwind concentration is Q (19?) . c HN ee) (203) ny! ua,ti "| BaP) ‘Turner (1970) suggests that Eq. (20.8) is also valid for the plume trapped under an elevated inversion (at height Z) if L is substituted for H and if is sufficiently far away from the source. This “sufficient distance is set equal to twiee the distance at which the edge of the ne begins to interact with the inversion layer. If the distance of first interaction is xz, then according to Turner (1970) o, at this di tance is given by the plume. Under these conditions, the equation that best predicts the downwind concentration is Q 19? kay Cee (208) 12 (2n)"?uo,H | 203 ‘Turner (1970) suggests that Eq. (20.8) is also valid for the plume trapped under an elevated inversion (at height L) if L is substituted for H and if x is sufficiently far away from the souree. This “sufficient? distance is set equal to twice the distance at which the edge of the plume begins to interact with the inversion layer. If the distance of first interaction is xz, then according to Turner (1970) o, at this dis- tance is given by 6, = 0.47(L-H) (20.9) from which xz can be determined from Figure 20.8. At the point of first interaction, no effect of multiple reflections has occurred. Clearly, Eq, (20.1) is valid up to this distance. For x > 2x1, Bq. (20.8) holds (with L substituted for H), Thus, we have an alternative to Eq, (20.7). For x < xz, use Eq. (20.1); forx > 2x;, use Eq. (20.8); and for xy, 20.4 Tall Stacks and Plume Rise Consider Bq. (20.1), written for the ground-level centerline con @o.1n tion (20.11) der the concentration onstrates that, with everything else constant, any point x should go down very quickly as /7 According to Wark, Warner, and Davis (1998), ipproximately as H? for many condition. This relationship is the reason for designing tall 8. The objec

You might also like