You are on page 1of 24
a ‘variations, missions trom short stacks are often exhibit large deviations from the Gax factions with local terrain and buildings, (Schnelle 197 not as well studied and an model owing to inter- ‘ables 20.5 and 20.6 6) give some data on tall stacks and other tall things. Design Procedures Modern-day design of tall stacks makes considerable use of com- puters, Computers are invaluable to the stack designer both for their bility to process large quantities of meteorological data and for their ability to quickly calculate expected 4 concentrations for a given set of design conditions. There are many different computer progr’ ns avail- Table 20.5 Heights of Tall Things ‘stacks Buildings ft tbdury, Ontario) 1250. ‘Burj Khalifa (Dubai) 2720 rg Power Mitchell Plant 208 CN Tower (Toronto) 1815 betiand Plant 1000 Empire Sate Building 1250 TVA Paradiso Plont 800 Eifel Tower 986 TVAGalltin Plant 500 Washington Monument 585 ee LPP IESE PEIN AINE ‘Raped fom Sexe 1076. Table 20.6 The Inco Copper Smelter tack, Sudbury, Ontario, Height 250 8 Diameter atop and at base 152 f (op), 16 f (base) Inside ameter atop 0” Stack gas velocity TANgINY, UVeR zUACIUE) UL /AY YE uMyLED aLUWeyeL, Le 4s duecessaLy account for plume rise, and therefore we discuss some of the better- known methods in the next few paragraphs. Several published accounts compare plume rise models with each other and with actual observations (Carson and Moses 1969; Thomas, Carpenter, and Colbaugh 1970). We refer the interested reader to the literature for further details. 10,000 1000 2 i E 8 S100 lw 4 10 1 70 Too 1000 Que VA ‘Moses and Strom riggs jlland Figure 20.13 Comparisons of various plume rise models. (Adapted from Briggs, 1975) 676 Chapter Twenty Atmospheric Dispersion Modeling One well-known plume rise equation is the Holland (1953) for mula, which at one time was recommended for use by the U.S. EPA. ‘The equation is: an-2ilrs sean 01» where v, =stack gas velocity, m/s u= mean wind speed at stack he a, P. 1, = stack gas temperature, K tack inner diameter, m atmospherie pressure, mb 7, = atmospheric temperature, K ‘The last term of the Holland formula is directly related to heat emis- sion rate (Wark et al, 1998), so Bq, (20.12) could be written as plate , 250 (20.13) whore Qj = heat emissions rate, MW "The Holland formula as depicted by Eq, (20.12) or (20.18) is valid for neutral conditions, and the plume rise obtained by it should be cor: roeted for other than neutral stability (Holland 1953). Suggested mul- Liplcative correction factors are 1.1 or 1.2 for Class B or A stabilit ‘and 0.9 oF 0.8 for Class B or F stability, respectively Another model with reasonable performance is the modified Con ‘T, = atmospheric temperature, K ‘The last term of the Holland formula is directly related to he sion rate (Wark et al, 1998), so Eq, (20.12) could be written as emis- Me dy 9.6Qy (20.18) where Qyy = heat emissions rate, MW ‘The Holland formula as depicted by Eq. (20.12) or (20.13) is valid for neutral conditions, and the plume rise obtained by it should be cor reeled for other than neutral stability (Holland 1953). Suggested mul: tiplicative correction factors are 1.1 or 1.2 for Class B or A stability and 0.9 or 0.8 for Class E or F stability, respectively. Another model with reasonable performance is the modified Con- awe formula (Thomas, Carponter, and Colbaugh 1970): ana 202 2(@u) (20.14) ‘The Briggs plume rise model (1969, 1970, 1972) is currently recom. mended by the U.S. EPA. The Briggs model is used in all EPA com- puter programs and appears to be better for large thermally dominated plumes. Briggs recognized that oven after a plume was bent over by tthe wind it continued to rise, owing to its thermal buoyaney. Thus, his equations predict 4h as a funetion of a buoyaney flux term, Fy (whic! 's usually dominated by thermal buoyaney), wind speed, and distance downwind, After a “long enough” travel time (or distanee downwind! the plume reaches its final rise, Different equations are used in the Briggs model, depending on atmospheric stability, First, for neutral or unstable conditions, the downwind distance to the point of final plume rise is x, where 7, = almospheric temperature, K ‘The last term of the Holland formula is directly related to heat emis- sion rate (Wark ot al, 1998), 50 Eq, (20.12) could be written a8 9.69 (20.13) w where Qu heat emissions rate, MW ‘The Holland formula as depicted by Eq. (20.12) or (20.13) is valid for neutral conditions, and the plume rise obtained by it should be eor- rected for other than neutral stability (Holland 1963). Suggested mul: tiplicative correction factors are 1.1 or 1.2 for Class B or A stability and 0.9 oF 0.8 for Class I or F stability, respectively Another model with reasonable performance is the modified Com- cawe formula (Thomas, Carpenter, and Colbaugh 1970): 01.2(@y)“ (20.14) =a i ‘The Briggs plume ris mended by the U ‘model (1969, 1970, 1972) is currently recom- ‘S. BPA, The Briggs model is used in all EPA com- puter programs and appears to be better for large thermally dominated plumes. Briggs recognized that even after a plume was bent over by the wind it continued to rise, owing to its thermal buoyancy. Thus, his equations predict Ah as a fumetion of a buayaney ux term, Fy (which is usually dominated by thermal buoyaney), wind speed, and distance downwind, After a “long enough” travel time (or distance downwind), the plume reaches its final rise. Different equations are used in the Briggs model, depending on atmospheric stability. First, for neutral or unstable conditions, the downwind distance to the point of final plume rise is xy, where T, ‘The last term of the Holland formula is directly related to heat emis sion rate (Wark et al, 1998), so Eq. (20.12) could be written as atmospheric temperature, K Bed, , 9.60 20.18) Ah =1.5: where Qj = heat emissions rate, MW ‘The Holland formula as depicted by Bq. (20.12) or (20,13) is valid for neutral conditions, and the plume rise obtained by it should be eor- rected for other than neutral stability (Holland 1953). Suggested mul- tiplieative correction factors are 1.1 or 1.2 for Class B or A stability and 0.9 or 0.8 for Class B or F stability, respectively. Another model with reagonable performance is the modified Con- cawe formula (Thomas, Carpenter, and Colbaugh 1970) 444 nth (20.19) ine Briggs plume rise model (1969, 1970, 1972) is currently recom- mended by the U.S. EPA. The Briggs model is used in all EPA com- puter programs and appears to be better for large thermally dominated plumes. Briggs recognized that oven after a plume was bent over by the wind it continued to rise, owing to its thermal buoyaney. Thus, his equations predict AA as a function of @ buoyaney flux term, Fp (which 's usually dominated by thermal buoyancy), wind speed, and distance ind. After a “long enough” travel time (or distance downwind) ches its final rise. used in the Briggs model, depending on neutral or unstable eonditions, the ‘of final plume rise is x, whore down: the plume Different atmosphi downwind distance to the pe Section 20,4 Tall Stacks and Plume Rise 677 x75 119(Fp * for Fy > 55 m4/sec! (20.15) or © 5=49(Fp)°® for My <55 m'/sec? (20.16) The plume rise (as a function of downwind distance, x) is caleulated by 16(y)* (s7)?° fora 2 xy 20.17) or 13 a= 2S)” oH) fore

rrrrrrrr Example 20.4 rless of wind speed. For Class D stability, calculate the final plume rise from a power plant stack, given the following information: a MW, = 28.9 eemol Use the Holland formula, the modified Coneawe formula, and the Bri Quy = 40,000 kis ss method, Solution Using Eq. (20.13) for the Holland formula, and reeognizing that 40,000 klfs = 40 MW, 1.5(20)5 , 9.6(40) LELTLS, $9 m (Holland) Brom Eq. (20.18, y= 40,000 8.9 gmal Use the Holland formula, the modified Coneawo formula, and the Briggs method. Solution Using Eq, (20.13) for tho Holland formula, and recognizing that 40,000 kd/s = 40 MW, 1.5(20)6 , 9.6(40) 6 6 4h = 89 m (Holland) a vom Eq, 20.14), 101.2(40)44 1 =150 m (Coneawe) "To use the Briggs method, first calculate the buoyaney flux. Using Ba, (20.19) with the first term equal to zero, 1013) s9( 15 ]40 =200.6 Next, caleulate 2p from Bq. (20.18), sinee F Fy xy =119(380,0) 254 m Section 20.4 Tall Stacks and Plume Rise 679. Finally, since the final plume rise is desired, use Eq. (20,17): 0.667 1.61 Y 0.333. ha 1808606)" (254) Ah = 221 m (Briggs) Notice the large differences among predicted plume rises using the three methods. For a stack with a physical height of 100 m, the effee- tive height obtained via the Briggs method would be 70% greater than that obtained via the Holland formula. i 444044 Critical Wind Speed Consider Eq. (20.11) solved specifically at max fr Onay- It is obvi- ous that the pre-exponential term is inversely proportional to x. But in the previous section, it was shown that Ah (and thus H) is inversely. proportional to w also. ‘Thus, the exponential term increases with increasing u, while the pre-exponential term decreases. The result is that a maximum Cynax exists at some wind speed, Ue, called the critical wind speed; that is, if we solve for Cinqx for various wind speeds and plot the values of Cinax versus wind speed, the plot would trace a smooth curve passing through a maximum. The critical wind speed depends on the stability and the specific plume rise model used. It is best determined through use of a computer. Pr rrrrrrr Example 20.5 4 ‘ous tnav tne pre-exponential term is inversely proportional to u. pucin the previous section, it was shown that A (and thus 22) is inversely proportional to u also, ‘Thus, the exponential term increases with increasing u, while the pre-exponential term decreases. ‘The result is that a maximum Cyyax exists at some wind speed, ue, called the critical wind speed; that is, if we solve for Cyyx for various wind speeds and plot the values of Cay versus wind speed, the plot would trace a ‘smooth curve passing through a maximum. The critical wind speed depends on the stability and the specific plume rise model used. It is best determined through use of a computer. >rrrr rrr Example 20.5, Consider the following data: Q = 110 g/s, class = D, h = 100 m, MW, 28.9, Py = 1000 mb, and Qyy = 10 MW. Using the Briggs plume rise equation, calculate Cj a8 a funetion of wind speed and find the crit cal wind speed, the maximum Cinq, and the distance at w Solution h it oceurs. ‘A computer program that corrected wind speed to stack height and used the Briggs plume rise equation was used to solve this prob- Jem. Various surface wind speeds were chosen. The results are pre sented in the table on tho following page. From the tabulated results given in the table, the eritieal wind speod at stack height for Class D stability lies between 8.9 and 9.7 m/s and results in a Cyyqq 0f 32 ng/m® ‘at distances between 6 and 6.5 km. 680 Chapter Twenty Atmospheric Dispersion Modeling Uys mis Am 20 30 40 203 50 182 60 169 790 159 80 152 100 1a 14444444 Other Stack Design Considerations As air blows past a structure, a low-pressure region forms behind the structure, rosulting in recirculating eddies and wake formation, In addition, if the structure is Inrge, disturbances in the flow system are experienced above and on either side of the structure. Typical distur= Dances from a building and from the stack itself are pictured in Figur 20.14. Those offeets may result in plume downwash unless the ‘designer takes steps to eompensate for them. The rogion of turbulence owing to buildings may extend to nearly twice the building height hy and downwind 5 to 10 times hy (Turner 1970), A well-known rule of thumb for stack designers is to make the stack height at least 2.6 times the height of the highest building near the stack Plume downwash behind a stack ean 1 Section 20.4 Tall Stacks and Plume Rise 681. Eddies (b) Figure 20.14 Typical flow disturbances owing to (a) a building and (b) a stack. Se qx La. >> rrr rrrr Example 20.6 A source emits 20.0 g/s of pollutant contained in 6000 m/min of stack gases (at P= 1 atm and 7’ = 390 K), The maximum expected sus- tained wind speed at stack hoight is 15 nvs. The heat emission rate is ‘TMW, For a wind speod of 8 mis (at all heights) and Class D stability, design a stack (specify the height and diameter) that will prevent ‘ground-level concentrations from exceeding 50 jg/m? anywhere down- ‘nd, Use the Holland formula for plume tise Solution ‘The inner stack diameter is calculated first, The stack gas velocity is set by p= LO x 16 mis = 22.5 mis From o,, the stack diameter is obtained by the eontinui Q=0,A, (f(a aan =a 225 For the conditions of this problem, (G2) = BO), -2.0(10) m= (2 Jw.” 20.0(10)° From Figure 20.9 at the interscetion ofthis value of Cu/@ jax and the Class D stability line, values for HT and yay are found; H=68m 17 kam + a,-(22) (4) 800060)!" me TOSI =2.4 m For the conditions of this problem, ( 3) __(60)(8) @ Jinax 20.0(10)° From Figure 20.9 at the intersection of thi 20.9 3 is value of (Ci Class D stability line, values for H{ and max 8° TUN H=68m =1.7km and the Xmax From the Holland formula, we have 1.5(22.5)(2.4, Mig PIES: If )., 28(7)= 185 nm ‘Thus, h2H-Ah = 49.5 m 0m tall and with a top. inner diameter of 2.4m. Specify a stack 51 adda’ aad ke Stack Costs i stacks are arbitrarily classified as fot) or tall (ean exc mating purposes, eed 1000 feet). A For cost-esti in about 120 fe either short (less thar Section 20.4 Tall Stacks and Plume Rise 683 short stack is essentially a vertical duet supported in some manner (either by its own structure, by an external structure, or with guy wires). It may be made of steel, FRP, or brick, and may be lined or unlined (depending on gas temperature, moisture content, and chemi- cal properties). A tall stack is generally free-standing, and requires much more attention to the structural aspects of design, since it must be strong enough to stand against sometimes quite strong wind load- ings. Tall stacks generally have other components such as strong foun- dations, ladders, sampling platforms, access doors, and even aircraft warning lights, ‘Tall stacks are quite a bit bigger at the base than at the top (e.g. the INCO smelter stack—see Table 20.6). The conts of short stacks depend primarily on three factors: stack height, stack diameter, and material of construction. ‘The U.S. EPA hhas developed a simple cost estimating equation as follows: C=ab? (20.23) where j C= purchased cost:in 1998 dollars, $/t of stack height De=stack diameter, in. a, b= parameters given in ‘Table 20.7 ir additional components, tall stacks are much more 1 thousand dollars, tall ill fa tall stack is more ks may run several million dol ol e ae abe than that of @ short stack. However, Vatavuk gees ad data from a well-known stack constructor and developed the Tring cost equation for tall stacks: Because of the expensive. While s where C= purchased cost in 1998 dollars, $/ft of stack height D =stack diameter, in. a, b= parameters given in Table 20.7 Because of their additional components, tall stacks are much more expensive. While short sta .cks may cost several thousand dollars, tall ‘stacks may run several million dollars, The cost of a tall stack is more site speeifie than that of a short stack. However, Vatavuk (1990) gath- fered data from a well-known stack constructor and developed the fol- lowing cost equation for tall stacks: iH (20.24) ‘Fable 20.7 Cost Estimating Parameters for Short Stacks Equation Paramotor’ ‘Applicable Range Material a Dan. H, foot pvet 12-96 <10 Plate ~ CS, coated? 6-04 20-100 Plate 904 SS" 64 20-100 sheet -Galv CS* 836 275 ‘Sheet —304 SS* 3-38 £15 # with Equation 20:22 S tomnies stot plat 4 carbon sea hes, abenlod Surat, cocted wth ane eet of shop pant stiness . chow pe 308 ) ee a the purchased equipment cost, ‘lo get the total installed cost, multiply by a factor of about 1.5. On the other hand, for tall stacks, Bq. (20.24) gives the TIC directly. Whether Eq. (20.23) or Hq. (20.24) is used to estimate stack cost, the result must be updated to current year dollars using the methods described in Chapter 2. Table 20.8 Cost Estimating Parameters for Tall Stacks! Liner Type Stack Diameter, ft Parameters? a b Steel? 15 0.0120 0.811 20 0.0108 0.851 30 0.0114 0.882 40 0.0137 0.885 Brick* 15 0.00602 0.952 20 0.00562 0.984 30 0.00551 1.027 40 0.00633 1,036 1 Caution: do not extrapolate outside height range (200 to 600 ft). 2 For use with Eq. (20.24). 3 Carbon-steel liner with 316 L stainless steel top section. 4 Acid-resistant firebrick liner. published in Vatavuk, 1990. ‘Source: Based on actual project data from Zum Constructors, Inc, > 20.5 Computer Programs for Dispersion Modeling (Point Sources) modeling is an attempt to represent by ved phenomena of pollut- AAEM acon ‘Atmospheric dispersion mathematical or statistical means the obse Section 20.6 Mobile Sources and Line Source Models 687 2 ees OZ d. Qnus,o, iy Note: Because of the increased turbulence caused by cars, the 9, and 6, functions must be modified when modeling dispersion from road- ways. It is beyond the scope of this text to discuss that modification in detail. For further reading, see Benson (1979), Since o, and o, are constant with respect to y, Eq. (20.25) can be integrated over the FLLS and written as follows: (20.26) where ‘The reason for leavin, ig the quantity 2x0, become obvious shortly. Define out of the constant K will a variable B such that (20.28) and substitute into Eq, (20.26) for y, The result is 1K pp BY) C= Fra In 0%”|-2 lao, (20.29) ‘The reason for leaving the quantity 2noy out of the constant K will become obvious shortly. Define a variable B such that 20.28) and substitute into Eq (20.26) fory. he result is 1 1A 5A) BAe on B lupo, (20.29) , } ‘exp| —— |d (20.30) eo =F Jam » o-£(e,-0) ens wassian distribution function evaluated at By where Gy, Gy = th and By, respectively ‘The Gaussian distribution function is discussed briefly and tabu- lated in Appendix C. "Thus, Bq. (20.81) ean bo applied to solve the FLLS problem. Ifa roadway is modeled by several FLLS, then the eon. tributions of all FLLS to the receptor are summed to obtain a total tomeenteation, This method is illustrated in the next example 688 Chapter Twenty Atmospheric Dispersion Modeling >>r>rrrrr> Example 20.7 Given the geometry of a road segment and receptor shown in Fig- ‘ure 20.17, and the data below, ealeulato the expected ground-level CO concentration at the receptor. Use the two subsegments shown to get ‘ovo equivalent finite length line sourees, and assume a constant emis sion rate along each FLL, CO emission factor = 1.0 m/s, H= 0m, and tr that c, = 20 m, and o. Mmats=67.5 m, 15 g/km, Class is D, w fic count is 6000 vehiclos per hour. Assume 12 mat x= 50m, and that o, Solution First, a uniform emission rate on the road is ealeulated: (000m “Te 4 = 25,000 ums Noxt, equivalent FL are dravin throu and FLL m. The sogments that are perpendicular to the wind midpoints of the two road segments. From the th of FLLS; is 60 m distance d is 70/2 sin 80° = 17.5, sivalent emission rate on FLLSy is 8,000 70 n/60.6 m = 28,867 pgin-s 1000m) 1g = 25,000 ne/m-s Next, equivalent FLLS segments that are perpendicular to the wind ‘are drayin through the midpoints of the bwo road segments, From the ‘The length of FLLS, is 50 m ceometry, Figure 20.18 is prepared. ‘= 60.6 m. The distance d is 70/2 sin 30° = 17.5 given g and FLLS, is 70 eos 30° 'm. The equivalent emission rate on FLLSy is (q' = 25,000 x 70 m/60,6 m = 28,867 ng/m-s Figure 20.47 Geometry for Example 20.7. Figure 20.18 ‘Skeich to accompany Example 20.7. ‘The coordinates 91, ya, and yg (relative to the x-axis defined by the receptor) are 55.0, 5.0, ind 455.8, respectively. The x distances from FLLS, and FLLS, to the receptor are 50 m and 67.5 m, respectively. ‘At these distances, c,, and o,, are 20 and 22 m, respectively, and 6, and 0, are 12 and 14 m, respectively. us consider FLLSy now: = B= oy, 20 4 Brom Bq, (20.21) forz = 0 and Hi -. 25,000 = 4167 Ky = gyrate 20)= 46 From Eq, (20.31) and the table in Appendix C the concentration at the receptor owing to FLLS} is = 82 fc(-0an -o(-20] 1604010003052 yen! N const or PLS

You might also like