ML17055C375
| ML17055C375 | |
| Person / Time | |
|---|---|
| Site: | Nine Mile Point  |
| Issue date: | 09/05/1986 |
| From: | Halitsky J JAMES HALITSKY |
| To: | |
| Shared Package | |
| ML17055C371 | List: |
| References | |
| NUDOCS 8609120406 | |
| Download: ML17055C375 (32) | |
Text
{{#Wiki_filter:ATTACH?IE'l<T 3 COttCEttTRATION COEFFICIENTS IN ATMOSPHERIC DISPERSION CALCULATIOttS By James Halitsky, Ph.D. ~ The concentration coefficient technique is a method for calculating the .concentration field in the atmosphere in the vicinity of a building when an airborne substance is released from or near the building. The operative parameter in the concentration coefficient K which, in the ma)ority of applications, is independent of substance release rate, wind speed, and building size, but is dependent on wind direction, building shape, source configuration, and receptor location. The prediction of real concentrations depends on one's ability to make an appropri-ate estimate of K and to transform it into a concentration estimate. This paper describes available. data sources for K, examines its nature, and illustrates its use by a case study. Atmospheric dispersion is a mixing process whereby airborne matter is spread over an ever-increasing volume of airspace by the turbulent motion of the atmosphere. A continuous release of matter into a steady wind produces a stationary (time-independent) plume, characterized by nonzero concertrations of the dispersed matter. This paper deals with the calculation of such concentratiors in plumes created from sources near building surfaces. Plumes of this type have concentration distributions different from the Gaussian distribution that exists in free-stream plumes lying well above the region of wind disturbance created by the building. The most accu-rate method for estimating concentrations in stationary plumes from building sources is the concentration ccefficient technique. A concentration coefficient is a nondimensional representation of a real concentration in the same serse that a pressure coefficient is a nondimensional representation of a real pres-sure. In both cases, the coefficient is found by dividing a measured quantity by an artificial reference quantity constructed from the field boundary conditions. For a pressure field, the reference quantity is the dynamic pressure. For a concentration field, the refer ence quantity is an artificial concentration C ref (amount/volume) created from the release rate Q f (amount/time) of pure matter, the mgan wind velocity U (length/time) at a designated location, and a characteristio area A (length ), producing C f = Q/AU For a concentration C (amount/volume) at a specified point x,y,z, the corresponding concentra-tion coefficient K (dimensionless) is (2) K a C/Cr f= CAU/Qo ~ J. Halitsky is a consultant in Environmental Meteorology, 122 N. Highland Place, Croton-on-Hudson, NX t0520. PSHtt~ ft AvD411 C'one . 19$ J 860'F120406 8b0905 PDR *DOCK 050004fO( A PDR
The usefulness of K derives from the fact that, in many cases, constant for a wide range of magnitudes of Q, A, and U. This rakes itit remains substantially possible to estimate C in the full-scale atmosphere from K obtaired in a wind tunnel rodel test, simply by using Equa-tion 2 with full-scale values of Qy Ay and U. The feasibility of the technique rests on the availability of published K data and the validity of applying the data to a Pull-scale confi-guration which does not resemble the model confLguration in detail. This paper contains ( 1) a review of the available data base for estimating K, (2) a sec-tion on the nature of K, (3) a case study to illustrate the application of the concentration coefficient technique to a complicated building, and (4) an evaluation of the accuracy of the technique by comparison of the estirated concentrations with concentrations measured in a wind tunnel model test of the study prototype. K Ls usually reported in the literature as K isopleths in the airspace surrounding a building or group of buildings. Only three fairly comprehensive studies have been made on simple geometric shapes, but these are important because such shapes can be construed to resemble por-tions of larger, more complicated structures. The experiments that produced the K isopleths were conducted with wind tunnel models; the full reports on testing procedures, data reduction, and data interpretation are given in laboratory prospect reports References 1, 2 and 3. Abridgements of References 1 and 2 appear Ln the open literature as References 4 and 5. An abrLdgement of Reference 3 appears in Reference 6. References 7 and 8 present an up-to-date ( 1980) compilation of research data on flow and diffusLon near buildings and contain some of the K isopleth drawings Ln'References 1 and 3. Lrmu~a An example of a K Lsopleth drawing, abridged from Reference 1, is shown Ln Figure 1. The Lsopleths are drawn in the plane of each visible building surface and Ln the airspace in sec-tions above and to the side of'he building. Each isopleth Ls identified by a value of K rang-ing from zero at some distance from the building to a maximum at the exhaust port. Each iso-pleth line is, in fact, the intersection oi a constant K surface with a building surface or a section in space. A mental reconstruction will show that the space around the building is occupied by a continuous field of K, made visible by discrete constant K surfaces. The K Lsopleth surfaces and lines were created by interpolating curves through an array of K data points which had been obtained by transforming measurements of C Ln a wind tunnel test to corresponding values of K by Equation 2 and plotting them Ln a space created by dividing all real lengths by the buildiqg height H. Therefore each data point represented a nondimensional-ized concentration at a nondimensionalixed location, i.e., C (x,y,z,) - K (x/Hyy/Hyx/H)g (3) and the constant of proportionality is C reff of Equation 1. In establishing C ef and it is necessary to adopt some convention for the desig ations or Qi Invariably, the source strength Q is the flow rate of pure contaminant passing A, U. through the exhaust port cross seotion. Hore flexibility is available for A and U ~ It Ls preferable, but not mandatory, that A be associated with a distinctive feature oi the flow field around the building. This feature Ls a local zone of toroidal circulation, called a cavity, lying within a large disturbed flow xone called a wake. The cavity and wake originate, and are coincident, at the upwind edges of the building, but the cavity is finite in length and maxLmum cross section, while the wake cross section grows continuously with distance downwind until the wake disappears as free stream kinetic energy diffuses into it.
Cavity flow controls dispersion near the building because the contaminant is usually discharged within the cavity or sufficiently close outside to disperse in a flow fLeld that must conform with the cavity shape. The building feature that contributes most to the creation of the cavity flow is the building frontal area progected on a plane normal to the wind. References and 3 both employ the frontal area for A but with a difference that may be signi-ficant if 1 the building is long and narrow. 'The A in Reference 1 is the frontal area of the largest side of the building; it is invariant with wind direction. The A in Reference 3 is the frontal area pro)ected on a plane normal to the wind; it varies with wind direction. The latter is more closely related to cavity sixe, but the former Ls simpler to use in calcula-tions. Both are acceptable, but the Lsopleths in each reference were derived with the desig-nated A and should remain associated with it in subsequent applLcations. Two candidates for reference velocity are the exhaust stream velocity V and the wind velocity U. Both, acting together, defire the total flow field, but the former dominates in the region near the exhaust port while the latter is more important elsewhere in the cavity. The latter is more commonly used because receptors near buildings will be in regions of low concentration away from the port vicinity. References 1 and 3 both use wind velocity at roof height for U. The heLght specifLcation Ls not important Ln Reference 1 because the wind had a uniform mean velocity profile. Zt is significant in Reference 3 because the profile was of the boundary layer type, L.e., velocity increasing with height from mero at the ground as in the natural atmosphere. Use of frontal area for A and roof wind velocity for U is suitable for isolated structures as in References 1 and 3. Practicality dictates other choices in other situations. For exam-ple, in reporting full-scale tests of dispersion at a nuclear reactor complex in Reference 10, A was defined as the frontal are of the reactor contair~ent structure alone, although many other large buildings Ln the complex also contributed to cavity formation, and U was defired as the wind velocity at the 6 m (19.7 ft) level on a tower located 600 m (1,969 ft) upwind of the co plex. K values derived from this reference must be ad)usted for use with A and U specified Ln other building arrangements. The only other refererce parameter is the length used for nondimensionalixing real dis-ances. Building height H was used for the block buildings in References 1 and 3. The build-ing in Reference 2 was a half-sphere atop a vertical cylinder; the reference length was selected to be the diameter rather than the height. This convention, also, must be retaired in locating K values from Reference 2. The constancy of K over a range of scales depends on invariance of the normalized flow fLeld in which the dispersion takes place. Such invariance occurs, according to the hydro-dynamio equations of motion and dispersion, when the configuration is invariant and a minimum Reynolds Number is observed. A configuration Ls a statement of normalized boundary conditions for a specific flow and dispersion field. The fLeld under consLderation here is the atrosphere in turbulent motion over the building and surrounding terrain. Although the field Ls effectively infinite in extent upward and horixontally outward from the building, it is convenLent and sufficiently accurate to consider only that portion in an imaginary box on the ground enclosir~ the build-ing, The walls and roof of the box are set only far enough away from the building to provide wind properties in their planes essentially the same as if the building were absent. tance of three building heights Ls usually required in the lateral, vertical, and upwind direc-A dis-tions; a larger distance is required in the downwind direction to accommodate the slow decay of wind disturbances created by the building. The configuration has three components. The geometric confLguration Ls the shape of the faces of the box, fLve of which are orthogonal imaginary planes and one, the bottom, is an irregular solid surface conforming to the terrain and building exterior contours, continuous except for the exhaust and intake ports. The dynamic configuration is the distribution of velocities along the box faces. (Density and temperature are ignored in the present context because differentials between building and atmospheric air are too small to make signLficant changes from the flow patterns that would exist under isothermal conditions.) The source con-figuration is the distribution of concentration across the exhaust port. Two configurations are said to be the same Lf the properties of one convert into the pro-perties of the other when multiplied by a single constant. The constant is the ratio of magni-tudes of a charaoteristio property. For geometric similarity, it is the ratio of reference
lengths. For dyr~ic similarity, it is the ratio of reference wind velocities. For source sLmilarity, it is the ratio of reference source concentrations, which is proportLonal to the ratio of source strengths when geometric and dynamLc similarity are present. The minimum Reynolds Number, formed from the reference length, the reference velocity, and the density, restricts the amount of turbulent energy converted to heat by molecular interac-tion to a very small value, thereby preserving the turbulent character of the field. This cri-terion is important in setting wind tunnel test conditions because the small model size creates a small Reynolds Number. It is automatically observed when model-generated K isopleths are used for estimating full-scale concentrations because the full-scale flow field will have a higher Reynolds Number than that i,n the test. In practice, exact similarity of configurations between model and full scale is not possi-ble or necessary, provided that substantial similarity is achieved in maJor features. For example, in model testing of a given prototype building, miror surface irregularities are omit-ted and larger protrusions, recesses, and even other buildings located outside the exhaust-intake path are replicated only in crude block form. 1ihen using model-derived isopleths to estimate full-scale concentrations for another building configuration, greater deviation from similarity is to be expected; however, considerable deviation can be tolerated without too great a penalty in reduced accuracy. As an aid in estimating K when exact similarity is absent, itoutside is useful to be aware of two regions where K takes on characteristic values (apart from K = 0 the plume). f I One such region is the vicinity of the exhaust port. The exhaust mixtur e crosses the plane of the port through exhaust area A e with uniform velocity V e and concentration C e, carry-ing contaminant flow Q, all related by C =Q/AV (4) The corresponding K value Ls found by setting C in Equation 2 equal to C e in Equati,on 4 to obtain K = (A/A )(U/V ). (5) 2 For example, in Figure 1 the model building was a 0.38 m (15 in) cube with A ~ 0.145 m2 (1.56 ft )2 the exhaust port was a 0.0127 m (0.5 in) demeter circle with A = 0.000127 m (0.00136 ft ), and 0 and V were each equal to 1.22 m-s (4 fps), yielding K = 1,146. The same configuration for a Pull-scale building in the atmosphere would have thR same K, since A would be scaled by the same factor as A and U by the same factor as V . It should be noted that K is a function of the building/port area ratio and wind/exhaust velocity ratLo. If either oP these should changers K would also change and the FLgure iso-pleths would no longer, be exactly transferable to the Pull-scale building. However, if1 the change still resulted in complete capture of the contaminant in the cavity, the changes Ln the isopleth pattern would be localized to the exhaust vicinity. Another region where K takes on a special value is at the downwind end of the cavity. Despite a small vari. ation of concentration with height, which is responsive to exhaust port conditions, a fairly uniform average concentration Cw is created in the swirly, hLghly-turbulent cavity flow. C is given by C = Q/A 0 (6) where A is the maximum cross-section area of the wake normal to the wind direction and U is the average wind velocity in the contaminated region Just downwind of the cavity, and K fol-lows by combining Equation 6 with Equation 2: K a (A/A )(U/U ), (7)
Heasurements show that A 2A and an average Uw U/3; Equation 7 then yields Kw = 1.5. That this is a realistic viue may be seen in Figure 1; K is about 1.5 to 2 over the downwind tace of the cube where the exposure is mainly to well-mixed cavity concentrations in the return flow. Although Figure 1 is specific to a cubical building, similar K patterns appear with block buildings having different proportions but all having roof exhausts. For example, three sets of K isopleths for a block building having sides in the ratio 1:3:3 and with difterent faces presented to the wind are shown in Reference 4, Figures 20, 21, and 22. All have K values of about 1.5 at the lee face, although smaller values appear in the lower half of the tall narrow building because of strong horirontal infusion of fresh air near the ground. (Note that the Ke (= K max ) values in Figures 20 and 21 are incorrect; they should be interchanged.) In Reference 6, for block buildings in a boundary layer, isopleths at the downwind wall have about the same'verage value of 1.5 as in Reference 4, although the range (trom large at the roof to small at the ground) is greater. The difference is due to the difference in confi-gurations. The Reterence 6 buildings had very small exhaust ports and were immersed in a deep boundary layer. The wind stream separated at the upstream building edges but reattached to the roof and walls and separated again at the downwind edges to form a lee cavity that was smaller than the Reference 4 cavity, with Aw " A. This alone would double Cw and Kw . However, because the release was in the smooth flow, some of the contaminant diffused upward before reaching the cavity and was not recirculated to the lee wall. The loss of contaminant and the reduced Aw contributed in opposite ways to fortuitously producing an average K that was about the same in both tests. Round buildings, such as nuclear reactor contairment structures, produce smaller cavities than sharp-edged buildings; therefore, K should be larger, according to Equation 7. Reference 5, Figure 5.29c, shows K averaging about 3 at the end ot the cavity (about 2.25 diameters downwind of the building center. The prevalent appearance ot an average K ot 1.5 at the lee face of a sharp-edged building is a powerful generalixation that may be used in design of wall and ground intakes when the exhaust is on the roof, but a few words of caution are warranted. The existence of K 1.5> stemming from Equation 7, implies that all of the released. contaminant is trapped in the wake and tlows downwind through A . This is true for exhausts whose jet velocity, diameter, and elevation are insufficient to thrust the contaminant through the cavity boundary. Figure 1 is an example of this condition. However, large-diameter high-velocity jets from stub stacks on roof-rounted fans often have sufficient momentum to penetrate the boundary, allowing some of the contaminant to escape and leaving the balance to create a reduced Q, say tQ, which creates The estimation of f is beyond the scope of this paper since it requires familiarity with t e interaction of jet plumes with roof cavity flow. Appendix A provides some comments on this subject. Roof dispersion patterns are controlled by two factors; the presence or absence of a roof cavity at the exhaust port and the strength of the exhaust jet. If a cavity is present and the jet is weak, the pattern is similar to Figure l. If a cavity is absent, os if there is a cav-ity and the jet is strong enough to penetrate the boundary, the pattern is that of a plume from a short stack. A roof cavity will be created whenever a sharp root edge is presented to the wind. It the edge is normal to the wind, the cavity will extend the entire width ot the root and part or all ot its length, depending upon the building proportions and the approach wind velocity profile. It the edge is at an angle to the wind, the cavity will cover a portion ot the roof contiguous to the edge, the sire of the coverage decreasing with greater departure ot the edge angle from normal. The portion of a roof not covered by a cavity may be considered as a region ot smooth flow in the direction, and at the velocity ot, the approach wind. Such regions occur in normal orientation downwind ot the cavity when the building proportions and approach wind velocity profile are such as to create flow reattachment. Guidelines tor estimating the reattachment region may be tound in References 7 and 9. Smooth flow regions also occur at the center of the root in orientations other than normal. Figure 2 (abridged from Reference 4) shows K isopleths tor a cube in 45 orientation; the smooth flow region lies between the K s 0 isopleths, and the cavities are at the lateral corners. The plume, which is well formed over the building, des-cends rapidly atter passing the downwind corner and is wholly captured in the cavity. The
average K on the lee face Ls about 3, owing to local downward diffusion of high plume concen-trations, which augment the more dLffuse concentrations Ln the return flow. In evaluati,ng a proposed design, the wind direction should be allowed to rotate 'through 360 and the locations of exhaust port and intake observed with respect to roof cavities. In any one direction, if both exhaust plume and intake are Ln the cavity, a K value at the Lntake should be selected frcm among the published K isopleths and adJusted for the difference between the design Ke and the published K . Note that this adJustment Ls greatest near the source, where K is dependent on A and 5 , and will not be necessary at the lee wall unless the frac-tion of 5 retaired in the cavity cPianges. If the exhaust plume and the intake are outside the cavity, dispersion should be calculated as in the open atmosphere. An ambiguity in the foregoing discussion is the criterion for establishirg when an exhaust plume Ls fully trapped in the cavity and when it can be considered to have escaped into the free stream. Appendix A offers some suggestions in this regard. An estimate was made recently of dLspersion of contaminated air released in several possible modes near the surface of the reactor enclosure building of a nuclear power plant. The esti>> mate was followed by a wind tunnel test of the same plant. The estimate and the test results provide an opportunity to illustrate how the K concept is applied Ln practice, to evaluate the predictive technique in gereral, and to demonstrate how seemingly small deviations between the conceptual and real confLgurations can produce significant changes in the concentration field. The facility has two reactors, designated Units 1 and 2, each in its own enclosure build-ing but served by a common control building and serving a common turbine building, all Joired to form the irregular building in the center of Figure 3. Two large natural draft cooling towers (N), two mechanical draft cooling towers (M), an electrical switchyard, and various buildings surround the central structure. The ventilation exhaust system for each unit has two internal pathways, only one of which will be Ln use at a given time. Each pathway terminates in a small louvered penthouse, desig-nated inboard vent (IV) or outboard vent (OV), on the roof of the unit's auxiliary building. A design alternative under consideration was a stack release whose por t was at the elevatLon of the roof of the enclosure building. A fourth (accidental) release mode was seepage through the exterior walls of the enclosure building. The receptor was considered to be in the control room in the interior of the control building. The contaminant could enter the control room via fresh air intakes spanning the west wall of the control building at an elevation of 13.7 m (45 ft) above ground, or, if the intakes are closed, by infiltration through the roof of the control building. An estimate of concen-tration at the center oi the wall intake (B) and at the center of the porous roof area (D) was desired. The prototype wind condition was a natural boundary layer Ln neutral stability with U 1.52 m-s (5 fps) at anemometer height of 58.5 m ( 192 ft). This corresponds to a wind speed of 1 m s (3.3 fps) at the 10 m (32.8 'ft) elevation, a conventional assumption for nuclear reactor accident dispersion calculations. (The atmospheric stabLlity usually assumed in such calculations is strongly stable, with a consequent difference in the approach wind turbulence and mean velocity profile. However, the wind disturbance created by the building overwhelms the approach wind characteristics, making the assumption of neutral stability valid,) In selecting the appropriate set of K isopleths, it is necessary to evaluate the control-ling features of'he flow between source and receptor. Both IV and OV for each unit are located at the surface of a large composite structure consisting of the unit's combired enclo-sure and auxiliary buildings. The small louvered penthouses over the exhaust ports destroy momentum Ln the exhaust Jets, ensuring that the release is at the surface of the struc-any'pward ture. Receptors B and D are located about one structure diameter from its center. By
L l l( 0 l '
inspection, the critical wind directions (producing highest receptor concentrations) appear to be MSM for Unit 1 and NNE for Unit 2. The receptors are in the lee of the structure in these wind directions. The eft'ect of the four cooling towers, whose wakes and internally-generated external air circulations alter the approach wind characteristics in a complicated manner, are not considered in the estimate. The confLguratLon suggests that, Ln each wind direction, the receptors are at the bottom of a large building cavity which is contaminated by releases from small surface ports. The isolated containment structure tests of Reference 2 fulfillthese criteria. Horeover, Refer-ence 2 provides the only available set of isopleths in space downwind of a building. Figure 6 is a reproducti.on of one of several K-Lsopleth drawings in Refererce 2; it represents an upwind mid-height release location. The Reference 2 building, identified as the EBR-ZZ contairwent structure, is a circle in plan view. Circular approximations to the Unit 1 and Unit 2 composite structures are shown as dashed circles of 61.0 m (200 ft) diameter in Figures 4 and 5. (The orLentation of these fig-ures is such that the wind passes from left to right across the page.) The upper parts of the figures are elevation sections through the centers of the circles, with the EBR-IZ building shown in its correct propor tions. The placement of the EBR-IZ building was such as to match the ma)or dimensions of the composite structures Ln plan view and at the top in elevation. This resulted in an EBR-IZ base at 16.8 m (55 ft) below plant grade. The K Lsopleths in Reference 2 are presented as seven drawings, each corresponding to a small surface source at a different location: top; midheight upwind, side, and downwind. The appropriate set for use herein was chosen to provide matchi.ng source locations. For Unit 1 Ln a MSW wind (Figure 4), the vents clearly are at the upwind location but the height is uncer-tain. For Unit 2 in a NNE wind (Figure 5) the vents are midway between the side and dowrwind locations and, again, the heights are uncertain. To establish the effective vent height Ln proportion to the structure height, Lt Ls impor-tant to consider the effective grade in the region of the cavity. The grade establishes the frontal cross-section area A, which creates the cavity cross-section area A . Both these areas contribute to the establishment of the average K at the downwind end of t5e cavity from which the return cavity flew toward receptors B and D arises (see discussion in connection wi.th Equa-tions 6 and 7). The effective cavity grade is shown in Figures 4 and 5. It was calculated by an averagig algorithm that took into account roof elevations along the section line as well as at ~ 22.5 in axLmuth from the section 1Lne as representative of the average height of the cavity base. For Unit 1-MSW the effective cavity grade was 1).3 m (60 ft) above plant grade, and the EBR-II net A above cavity grade was 22017 m2 (21,$ 13 ft ). For Unit 2-NNE, the corresponding values were 7.6 m (25 ft) and 2,667 m (28,710 ft ). The height of the vents in proportion to the building height above cavity grade Ln Unit 1-MSM confLguration is shown in FLgure 6. Zt corresponds well to the midheight EBR-IZ source height. Similar agreement Ls obtained in the Unit 2-NNE configuration (not shown), although the vents are slightly above the source due to the lower effective grade. It remains to be established that the K isopleths for midheight sources in Reference 2 can be used in the present case in view of the violation of the requirement of geometric similar Lty, i.e., the buildLng height/building diameter ratios are different. The only argument I can offer is the observation previously made that the average K value of the lce wall for block buildings have height/width ratios of 1 and 1/3 is about 1.5 for both, and the distributions over the wall from top to bottom are similar (higher at the top, lower at the bottom; see Reference 4, Figures 20 and 21). By analogy, a vertical contraction of the EBR-II.Lsopleths in the same proportLon as the height contraction should be permitted. It was rot necessary to re-draw the isopleths to a contracted ver tical scale as long as the receptors were placed in the correct vertical relation in Figure 6. The placement of B and D horixontally in Figure 6 was done straight-forwardly by dividing real downwind and crosswind distances by the scaled EBR-IZ diameter to obtain the x/D and y/D coordinates. Vertically, both receptors were placed Ln the base plane, even though the proper tionate height procedure put B below grade and D above in the Unit 1-MSM configuration and both above in the Unit 2-NNE configuration. The rationale for this in the case of D was that iso-pleths in the ground plane do not change during vertical contraction of the field, and, since D
in the roof plane, it was more reasonable to leave it in the ground plare than to was a set itnopoint above a mathematical rigidity. Zn the case of B, a location below grade is meaningless since isopleths exist there. The presence of Unit in the cavity of Unit 2 in the NHE wind direction creates a complex 1 interference flow pattern whose effect on the isopleth pattern could not be predicted; it was ignored for the estimate. of Table The interpolated values of
- 1. The Unit 1-WSW K
't at B and D for IV and OV sources are given in the upper part values were obtained from Figure 6. The Unit 2-HNE values were obtaired from similar figures based on Reference 2, Figures 11 and 14 ranged from 2.5 to 3.4. At the roof (D), K ranged from 2.3 to 2.8.'
the wall (B), K values for the stack releases were not estimated because of uncertainty as to hcw much of the released Q would enter the cavity. Fractional capture was thought to be quite likely since the stack release elevation was 12.5 m (41 ft) higher than the vents, and the stacks were uncapped. K values for the seepage release are given in Table 2. They were obtained by averaging the K values at B and D from all seven of the drawings of Reference 2, with the side release drawings yielding two values, one for the receptors on the same side as the source and the other for the receptors on the opposite side. The rationale for this procedure is that a seepage release would occur over the entire exterior surface of the enclosure building. The averaging procedure over all sources was considered to be the best representation. The average K ranged from 5.9 to 10.1. The estimating procedures described above differ from those in the pre-test estimate in two respects. One is in the placement of D at the base instead of at the proportional eleva-tion; this produced only a minor change in the estimate. The other is the use of a nire-point average for the seepage release instead of a single release at the downwind midheight source. This change was made because the wind tunnel test showed higher values of K by a factor of 2 to 3 than were originally predicted. Hindsight sparked the realisation that use of a midheight source alone omitted the important base release contribution. The test was conducted in neutral stability on a 1/240 scale model with tunnel wind velocity U a 3.1 m-s (10 fps) at the 58 5 m (192 ft) elevation and all other velocities (source and cooling towers) doubled to maintain their correct relation to the wind velocity. The N towers were made operational with internal axial flow fans, drawing tunnel air in at their bases and discharging it through their tops. The M towers were made operational by an external compres-sor that provided the correct outflow, but the air source was outside the tunnel. The test program provided concentration measurements at a total of 42 taps in the west wall and roof of the control building for eight release configurations (IV, OV, stack, and seepage on two units) and 16 wind directions with M and N on and off (all combinations were not tested). Some tests were also done with Unit 2 removed. The test results were reporte), in accordance with conventional practice in nuclear plant evaluation, as full-scale CU/Q (m ). K was found by applying Equation 2 to obtain A(CU/Q)t (8) K t with A as given previously for the WSW and NNE directions. To provide K values for comparison with estimates, four wall taps (Nos. 15-18) were aver-aged to represent B, and 16 roof taps (Hos. 27-42) were averaged to represent D. The lower parts of Tables 1 and 2 show test values for B and D i,n the Unit 1-WSW and Unit 2-HHE confi-gurations.
In the Unit 2-NNE configuration, the maximum test R for a vent release occurred at D with N and M off; its value was 2.6. The estimate for the sere confLguration was 2.3. For the seepage release with N and M off, the test value was 11.8; the comparable estimate was 9.7. The agreement is good (although the estimating procedure was changed in retrospect, as dis-cussed previously). On the other hand, very poor agreement was found in the Unit 1-NSN configuration. The maximum test K for a vent release was 0.1, whereas the estimate was 2.8. The maximum test K for the seepage release was 16.6 at D with N on and M off; the estimate was 5.9. The magni-tudes of the discrepancies and the fact that they occurred in opposite directions warrants an attempt at explanation. In the Unit 2-NNE configuration, the vents were at the point of flow separation at the south end of the auxiliary building roof. This ensured complete descent of the release into the cavity, and the rear presence of Unit 1 in the lee of Unit 2 created a blockage that strengthened the cavity circulation. These two factors provided for complete capture of the release in the cavity flow, producing the expected value of K. In the Unit 1-NSN configuratLon, the vents were at the upwind corner of a dLagonally-oriented building and the plume developed ver tically upward Ln traversing the roof, thereby placing some of the effluent above the capture zone downwind of the enclosure building roof. Second, Unit 2 was Ln a position to deflect some of'he wind passirg around the west side of Unit into the cavity region. 1 Third, the turbine buLlding intercepted the downflow at the end of the cavity and turned Lt downwind instead of back to Unit 1. The combination of frac>> tional escape of the developing plume, wind infection by Unit 2, and downflow interception by the turbine building resulted in almost no portion of the release reaching the receptors. For the seepage release, the low-level wLnd flow pattern in the cavity was altered by the presence of the lee unit and the turbine room to produce a flushing action at D for the Unit 1-NSN release, therby creating larger values Ln the latter configuration. The IV releases produced higher K values than did the OV releases, indicating some loss of Q prior to entry of the plume into the cavity in the latter case. Therefore, the OV location is preferable. The stack produced signLfLcantly love! K values. The maximum was 0.7 at D for Unit 2-NNE, compared to 2.6 for the IV release. This Ls attributable to the larger escape fraction with the hLgher release point. Of considerable interest is the lower K values at wall intake B. The test maximum was 1.0 compared to 2.6 at the roof. This is attributable to the lateral Lnflow of uncontaminated wind along the ground, striking the wall and flowing upward, thereby preventing the contaminated cavity flow at D from descending to B. The effect of the cooling towers appeared in the NNE orientation, as expected from their upwind placement. Operation of N reduced the maximum K at D from 2.6 to 1.8, with similar reductions for the other vent and the stack. Operation of M reduced K from 2.6 to 0.2. These reductions are attributable to wind disturbances created by the cooling tower plumes. The cir culation Ln a transverse Jet plume is a pair of counter<<rotating helical vortices, up along the plume centerlLne and dern on each side, and longitudinally downwind. Since N and M straddle Unit 2 in the NNE orientation, each plume produces a downflow at the unit. In additLon, N pro-duces a flow toward the east near the ground, while M produces a flow toward the west. The helical circulations alter and displace the Unit 2 cavity so that the receptors are in regions of lower concentration than in an undisturbed NNE wind. The effect Ls more pronounced when the wind velocity is low, as in the test. The influence of M is stronger than that of N because ,the M plume is closer to the ground (exit ports at elevation 17.4 m [57 ft]) while the N plume is high (originating as elevation 121.9 m [400 ft]). The concentration coefficient technique is generally belLeved to be quito accurate f'r predict-ing full-scale prototype concentrations from scale-model test measurements since such tests are designed to provide exact similarity in the important shape and flow characteristics. It also performs well Ln providing estimates of K in cases of nonexact similarity, provLded that the release can be associated with an isolated, cavity-producing structure for whLch K isopleths are available.
The method is less accurate when the release point is near the cavity boundary, since small changes in distance from the building surface may create large changes in the fraction of the plume captured by the cavity. It is also less accurate when wind flow aberrations are introduced by nearby portions of the structure or by more distant structures, such as cooling towers, which generate persistent helical circulations for long distances. It seems unlikely that K isopleths for configurations having these variations will be available in the near future. The alternatives for the designer are to improve one's ability to extrapolate beyond the available data by study of the r eferences, or to be consoled by the observation that in the present tests the maximum increase above the estimated value was 41$ (K = 11.8 increasing to 16.6 for the seepage release in the Unit 2-NNE configuration). In dispersion calculations, a factor of < 2 is considered acceptable. 1 ~ Halitsky, J. 1963. "Gas diffusion near buildings." New York University, Department of Meteor. and Ocean. GSL Rep. 63-3, Feb. 1963. (Loan copy available from author for dupli-cation.)
- 2. Halitsky, J.; Golden, J.; Halpern, P.; and Wu, P. 1963. 'Wind tunrel tests of gas diffu-sion from a leak in the shell of a nuclear power reactor and from a nearby stack." New York University, Department of Meteor. and Ocean. GSL Rep. 63-2, April 1963. (Loan copy available from author for duplication.)
- 3. Wilson, D. J. 1976. "Contamination of building air intakes from nearby vents," Universi-ty of Alberta, Department of Mechanical Engineering, Rep. 1, May, 1976, Edmonton, Al'berta, Canada.
diffusion near buildings." ~ ~~~ig,
~~
- 4. Halitsky, J. 1963. "Gas Vol. 69, pp.
464-485. 5~ Halistky, J. 1968. "Gas diffusion near buildings." Qt)JU1KGlggy EtlmLY 39K.i D. H. Slade, ed., U.S. Atomic Energy Comm. Div. of Tec. Inf., pp. 221-255. (Available as TID-24190 from NTIS, U.S. Dept. of Comm., Springfield, VA 22161.)
- 6. Wilson, D. J. 1976. "Contamination of air intakes from roof exhaust vents."
JZR1UU~NUA, Vol. 82, No. 1, pp 1028-1038.
- 7. Hosker, R. P., Jr. 1984. "Flow and diffusion near obstacles."
U.S. Department of Energy, Office of Sci. and Tech. Inf., Chapter 7. (Available as DOE/TIC-27601 from NTIS, U.S. Dept. of Comm., Springfield, VA 22161.)
- 8. Hosker, R. P., Jr. 1981. "Methods for estimating wake flow and effluent dispersion near simple, block-like buildi.ngs." NOAA Env. Res. Labs., Air Resources Labs., Silver Spring, MD, Tech Mem. ERL ARL>>108.
- 9. ASHRAE. 1985. fiat~~
ing, Refrigerating,
~ ~ujnm~, Chapter 14. New and Air-Conditioning Engineers, Inc.
York: American Society of Heat-
- 10. Halitsky, J. 1966. "A method for estimating concentrations in transverse Jet plumes, "
Vol. 10, pp. 821-843 ~ The sponsor of the case study investigation has made the K estimates and wind tunnel test results available for this presentation in the interest of'issemination of scientific data, but has declined to be identified. The wind tunnel tests were conducted at the Fluid Dynamics and Diffusion Laboratory, Colorado State University, Fort Collins, CO 80523.
44 APPEHDIZ A Penetration of a Jet Plume Through a Cavity Boundary The following discussion is based on get plume properties presented in detail in Reference 10 Peretration of a cavity boundary by a get plume is a much under-investigated sub]cot. Plume rise increases with emission velocity ratio V /U and decreases with rate of infusion of cavity air into the exhaust Jet. In comparison with Pree stream conditions, the local cavity wind is slower and more turbulent. The slower velocity creates a higher V /U and induces 8 greater plume rise. The higher turbulence and wind shear produce more rapid infusion of cavity air into the Jet, producing less rise. The two effects tend to cancel each other. I know of no analytical procedure or experimental data to quantity these trends. I suggest separate that the plume rise be calculated as if the cavity were nonexistent and a estimate made of the cavity boundary. Tho plume would then be considered to have escaped the cavity if the calculated plume centerline lies at least o above the cavity boun<< if the centerline lies below, at all distances up to dary, and to have been completely captured the receptor. This is a crude compromise based on the concept that most of the plume will des-cend into the cavity due to the high cavity turbulence when the centerline lies along the cav-ity boundary, and most of the plume will escape if the centerline is 2a Z above the boundary.
TABLE 1 K Values for IV and OV Releases
~MZL Mall Roof'all
~~&K Roof upwind midheight 9 2.5 2.8 downwind midheight 11 3.8 2.5 opp. side midheight f 14 Average 2.3 inboard vent IV on off 0 0.1 0.3 1.8 inboard vent IV off off 1.0 2 '
inboard vent IV on on 0.6 0.2 inboard vent IV off on 0.5 0. I outboard vent OV on off 0 0.1 0~3 1 3 outboard vent OV off off 0 0.1 0.8 2.1 outboard vent OV on on outboard vent OV off on 0.2 0.4 stack on off 0 0.1 0 0.2 stack off off 0. I Oej stack on on TABLE 2 K Values for Seepage Releases
~~2L MnQ~
Mall Roof Mall Roof Suucm upwind Paso 8 20 6 7 20 upwind midheight 9 3 3 3 2 top 10 2 5 5 2 downwind midheight 11 3 4 4 2 downwind base 12 19 14 25 17 source side base 13 35 10 15 35 opp. side base 13 5 5 4 source side midheight 14 3 ~ 3 4 3 opp. side midheight 14 3 3 ? Average 10.1 5.9 7.9 9.7 on off 0 16.6 0.1 12.8 off off 1+7 11.8 on on 2.7 5~9
0 n 1 10 5 2 0 2 K~ i. 15 V/U 1 0 /H ~ 1/30 K ~ 1146 Figure ). K isopleths for a cube in normal orientation to the wind. (Abridged from Reference 4, Figure 16.) MIND 0 10 1 2 K ~ V /U ~ 1 0 /H ~ 1/30 a 1146 Figure 2. K isopleths for a cube at b5 orientation to the wind. (Abridged from Reference b, Figure 19.)
8 'NATURAL DRAFT SCALE - FT COOLING TOWER 2 O 2SO Sno N AUXILIARY BUILDING MECHANICAL DRAFT ENCLOSURE BUILDING COOLING TOWERS TURBINE BUILDING M ELECTRICAL SWITCHYARD UNIT 2 UNIT I STACK PLANT IV TRUE' OV N WIND NATURAL DRAFT COOLING TOWER I NNE N WIND Piguce 3. Plan view of building at case study site.
WSM MIND~ STACK EBR II ANEMON. VENTS EBR II 190+ 41 345
~
192 PLANT ANEMONE 137 B 45 60 I PLANT GRADE 100R + AVERAGE CAVITY GRADE EL EVATION SECTION A-A UNIT 2 ro )0 UNIT I LENGTHS IN FT Figure 4. Local configuration for releases from Unit 1 in a MSM wind.
C' t0
EBR II NNE MIND ~ STACK VENTS AVERAGE CAVITY GRADE I PLANT GRADE VERTICAL SECTION B-8 UNIT 2 UNIT I LENGTHS IN FT Figure 5. Local configuration for releaaes from Unit 2 in a NNE wind.
=}}