Plant Plumes Volume 1: Field Observations and Theoretical

Transcription

1 Particle Formation and Growth in Power Plant Plumes Volume 1: Field Observations and Theoretical Studies of the Evolution of Particles in the Plumes From Coal-Fired Electric Power Plants EA-3105, Volume 1 Research Project Final Report, May 1983 Prepared by UNIVERSITY OF WASHINGTON Cloud and Aerosol Research Group Atmospheric Sciences Department Seattle, Washington Principal Investigators P. V. Hobbs M. W. EItgroth D. A. Hegg L. F. Radke Prepared for Electric Power Research Institute 3412 Hillview Avenue Palo Alto, California EPRI Project Manager C. Hakkarinen Environmental Physics and Chemistry Program Energy Analysis and Environment Division

2 ORDERING INFORMATION Requests for copies of this report should be directed to Research Reports Center (RRC), Box 50490, Palo Alto, CA 94303, (415) There is no charge for reports requested by EPRI member utilities and affiliates, U.S. utility associations, U.S. government agencies (federal, state, and local), media, and foreign organizations with which EPRI has an information exchange agreement. On request, RRC will send a catalog of EPRI reports. Copyright (C) 1983 Electric Power Research Institute, Inc. All rights reserved. NOTICE This report was prepared by the organizaiion(s) named below as an account of work sponsored Oy the Electric Power Research Institute, Inc. (EPRI). Neither EPRI, members of EPRI. the organizalion(s) named below, nor any person acting on behalf of any of them: (a) makes any warranty, express or implied, with respect to the use of any information, apparatus, method, or process disclosed in this report or thai such use may not infringe privately owned rights; or (b) assumes any liabilities with respect to the use of, or for damages resulting from the use of, any information, apparatus, method, or process disclosed this report. Prepared by University of Washington Seattle, Washington

3 ABSTRACT A parallel field study and theoretical investigation of particle size distributions in the plumes of coal-fired power plants is described. The field studies, which were carried out on a plant in Washington State and one in New Mexico, show that the particles in the plumes fell into three distinct size modes characterized by geometric mean volume diameters between pm, pm and pm. All three modes played important roles in the coagulation of particles in the plumes. The concentrations of particles sometimes decreased with travel time in the center of a plume while increasing at the edges of the plume. The concentrations of particles of a certain size were often less in certain regions of a plume than in the ambient air. Measured gas-to-particle (g-to-p) conversion rates were on the order of 0.1% of SOn/hr for the Washington plant and 1% of SO/hr New Mexico plant. for the A 3-D numerical model which allows for the effects on the particle spectra of advection, diffusion, coagulation, gravitational settling, g-to-p conversion and interactions with the ambient air, is shown to be capable of reproducing many of the principal features of the observed particle spectra in the plumes. The model predicts maximum g-to-p conversion rates at the edges of the plumes where the predicted rates are comparable to the field measurements. Elsewhere in the plume the predicted g-to-p conversion rates are about a factor of ten below the field measurements. 111

4 EPRI PERSPECTIVE PROJECT DESCRIPTION Sulfur oxides and nitrogen oxides have been implicated as major contributors to visibility impairment and acidic precipitation in the United States. Formation of small particles that scatter light, producing "haze" and act as condensation nuclei for raindrops has been reported to occur in the plumes of fossil-fueled power plants as well as other locations. The studies described in this two-volume report involved theoretical and field investigations of the rates of formation and size distributions of particles in six coal-fired power plants in the midwestern and western United States. PROJECT OBJECTIVES The objectives of the project, which spanned some five years, were to measure particle size distributions in coal-fired power plant plumes at various locations and under a variety of meteorological conditions. In addition, a three-dimensional numerical model was developed and refined for simulating particle formation and growth in plumes. PROJECT RESULTS The measured rates of conversion of sulfur dioxide to particulate sulfate ranged from 0 to 5.7% per hour. However, most observations fell within a narrower range of 0.1 to 1.0% per hour. Reaction rates were all found to depend on travel time from the stack and ultraviolet light intensity. The PHOENIX model predictions of maximum gas-to-particle conversion rates that occur near the edges of the plume are in agreement with the measurements. Elsewhere in the plume, the model predictions were about a factor of 10 lower than the measurements. The PHOENIX model has been applied in recent years by other sponsors to the specific question of haze generation in western plumes. Charles Hakkarinen, Project Manager Energy Analysis and Environment Division

5 ACKNOWLEDGMENTS This work was supported by the Energy Analysis and Environmental Division of the Electric Power Research Institute under Contract RP330. We wish to express our thanks to Dr. C. Hakkarinen of EPRI for his help and interest, and to the Pacific Power and Light Company and the Arizona Public Service Company for their cooperation in this study. v-n

6 CONTENTS Section BACKGROUND AND SCOPE OF STUDY Background Scope of Study 2 POWER PLANTS STUDIED AND AIRBORNE INSTRUMENTATION Power Plants Airborne Instrumentation 3 DATA ANALYSIS TECHNIQUES Particle Spectra Gas-to-Particle Conversion 4 RESULTS OF FIELD MEASUREMENTS Particle Spectra Gas-to-Particle Conversion 5 A NUMERICAL MODEL FOR PLUMES Introduction Background Rates Description of the PHOENIX Model Trace Gas Section of Model Particle Section of Model 6 COMPARISON OF MODEL RESULTS WITH MEASUREMENTS Introduction Measurement-Model Comparisons Centralia; March 30, 1976 Central ia; April 28, 1976 Centralia; October 19, 1976 Centralia; September 22, 1977 Four Corners; June 23, 1977 Four Corners; June 24, 1977 Four Corners; June 25, 1977 Discussion ix

7 Section 7 SUMMARY AND CONCLUSIONS Field Measurements The PHOENIX Plume Model 8 REFERENCES

8 ILLUSTRATIONS 4-1 Fitted curves to particle number concentration distributions 4-2 Excess concentrations (dn/d log D) above ambient values of particles in the plume from the Centralia Power Plant on October 19, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Central ia Power Plant on March 16, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Four Corners Power Plant on June 23, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Centralia Power Plant on September 22, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Centralia Power Plant on March 30, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Central ia Power Plant on April 28, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Four Corners Power Plant on June 24, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Four Corners power plant on June 25, Excess concentrations (dn/d log D) above ambient values of particles in the plume of the Four Corners Power Plant on June 29, Schematic of the processes considered in the PHOENIX plume model 5-2 Schematic of particle interactions by coagulation considered in the PHOENIX plume model 6-1 Excess (i.e. above ambient) concentrations (dn/d log D) of particles in the plume of the Centralia power plant on March 30, 1976, as predicted by the PHOENIX plume model 6-2 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SO/hr) in the plume of the Centralia power plant on March 30, 1976, as predicted by the PHOENIX plume model 6-3 Comparisons of measurements with predicted values from the PHOENIX plume model of the Aitken nucleus concentrations and ight-scattering coefficients through plumes along horizontal traverses perpendicular to the direction of the wind. 6-4 Excess (i.e. above ambient) concentrations (dn/d log D; of particles in the plume of the Central ia power plant on April 28, 1976, as predicted by the PHOENIX plume model XI

9 6-5 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SOo/hr) in the plume of the Centralia power plant on April 28, 1976, as predicted by the PHOENIX plume model 6-6 Excess (i.e. above ambient) concentrations (dn/d log U) of particles in the plume of the Central ia power plant on October 19, 1976, as predicted by the PHOENIX plume model 6-7 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SO/hr) in the plume of the Central ia power plant on October 19, 1976, as predicted by the PHOENIX plume model 6-8 Excess (i.e. above ambient) concentrations (dn/d log D) of particles in the plume of the Central ia power plant on September 22, 1977, as predicted by the PHOENIX plume model 6-9 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SO/hr) in the plume of the Central ia power plant on September 22, 1977, as predicted by the PHOENIX plume model 6-10 Excess (i.e. above ambient) concentrations (dn/d log D) of particles in the plume of the Four Corners power plant on June 23, 1977, as predicted by the PHOENIX plume model 6-11 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SO/hr) in the plume of the Four Corners power plant on June 23, 1976, as predicted by the PHOENIX plume model 6-12 Excess (i.e. above ambient) concentrations (dn/d log D) of particles in the plume of the Four Corners Power plant on June 24, 1977, as predicted by the PHOENIX plume model 6-13 Homogeneous gas-to-particle conversion rates (in units of 0.1% of S0;?/hr) in the plume of the Four Corners power plant on June 24, 1977, as predicted by the PHOENIX plume model 6-14 Excess (i.e. above ambient) concentrations (dn/d log D) of uncharged particles in the plume of the Four Corners power plant on June 25, 1977, as predicted by the PHOENIX plume model 6-15 Excess (i.e. above ambient) concentrations (dn/d log D) of charged particles in the plume of the Four Corners power plant on June 25, 1977, as predicted by the PHOENIX plume model 6-16 Homogeneous gas-to-particle conversion rates (in units of 0.1% of SO/hr) in the plume of the Four Corners power plant on June 25, 1977, as predicted by the PHOENIX plume model 6-17 Total particle concentrations (in units of 10 cm" across the Centralia plume at a distance of 2 km downwind predicted the PHOENIX model for April 28, xn

10 TABLES Table 2-1 Electrical outputs of the power plants during the periods of measurements discussed in this report 2-2 Specifications on research instruments on the University of Washington s B-23 aircraft 4-1 Modal parameters for particle distributions in the plumes from the Centralia power plant and the ambient air 4-2 Modal parameters for particle distributions in the plumes from the Four Corners power plant and in the ambient air 4-3 Ranges of values for the geometric mean diameter D and its geometric standard deviation a p. for the three particle modes 4-4 Rates of change of the fraction (fy.) of the total volume of particles in mode i 4-5 Summary of gas-to-particle (g-to-p) conversion rates deduced from the field measurements 5-1 Trace gas reactions included in the PHOENIX plume model 5-2 Ranges of concentrations of trace gases in the ambient air used in the PHOENIX plume model runs discussed in this report 6-1 Particle input data for case studies discussed in this report xm

11 SUMMARY RESULTS OF THIS STUDY Gas-to-particle (g-to-p) conversion rates have been measured in the plumes of six coal-fired power plants situated in the West and Midwest of the United States. The conversion rate was estimated from measurements of the changes in the total volume of particles in the plume, the production rate of Aitken nuclei and three different particulate filter analysis techniques (flash vaporization. X-ray fluorescence and ion-exchange chromatography) Comparison of sulfate concentrations derived from X-ray fluorescence and ion-exchange chromatography showed fair agreement-indicating that most of the particulate sulfur is sulfate. Comparison of g-to-p conversion rates estimated from the changes in total particle volume with those derived from particulate filter data showed, in general that the major portion of the g-to-p conversion product was sulfate. However, in some instances, the g-to-p conversion rate based on total particle volume was higher than the rate based on particulate filter analysis; this suggests that conversion products other than sulfate may sometimes be formed in power plant plumes. One possible conversion product, other than sulfate, which has been postulated to be produced in power plant plumes, is nitrate. To evaluate this hypothesis some nitrate measurements were made in the plumes. The data suggest that nitrate formation, in general was of ittle importance during the flights in which measurements were obtained. An estimate of the NO -to-m trate conversion rate was made for one of the cases studied at the Big Brown (Texas) power plant and was found to be "0.4%/hr for distances of 4.8 to 43.2 km. The S0,,-to-su1fate conversion rates, measured simultaneously, ranged from 0.7 to 2.8%/hr and increased with travel time from the stack. The rates at which new particles were nucleated in the plumes were evaluated and the ratio of this nucleation rate to the rate of formation of new particle volume 2 5 was calculated. The ratio was found to range from 2.7 x 10 to 3.9 x 10 particles pm with a mean value of 7.4 x 10 +/-.2 x 10 particles pm This ratio is S-1

12 indicative of the fraction of g-to-p conversion product which forms new particles --the remainder of the conversion product condensing directly onto already existing particles. The large variation in this ratio, some of which is attributable to differing plant locale, suggests regional differences in the g-to-p conversion process. Results from the PHOENIX plume model suggest that it should be possible to predict this ratio on the basis of a modified version of the theory of McMurry (27) The PHOENIX model outputs suggest the importance of the above ratio in determining the light-scattering coefficient and visibility degradation. Further analysis was made of the relationship between the rates of formation of new particle surface area and new particle volume. The data were found to be in fair agreement with the theoretical relationship of McMurry and Friedlander (64) in which the particle surface area varies as the rate of formation of new particle volume raised to the 3/5 power. This suggests that the size distribution of the particles in the plume becomes self-preserving. The SOp-to-particulate sulfate conversion rates were found to range from 0 to 5.7%/hr--the higher rates generally occurring in the Southwest. This range is in agreement with previous studies conducted both in the Eastern United States and at the Four Corners power plant in New Mexico. The SO? g-to-p conversion rate was also found to depend on travel time from the stack and UV ight intensity. Both the physics and the chemistry of the g-to-p conversion process(es) suggest that the dominant (though not necessarily the sole) conversion mechanism in the plumes studied is the oxidation of SO? by OH radicals. A significant correlation (r 0.9) was found between the conversion rates and a parameter indicative of this reaction. The results of this study have suggested several areas in which further efforts might yield valuable information. While the data collected in this study suggest that the SOp-to-sulfate conversion process is predominantly SO? oxidation by OH radicals, this has not been conclusively demonstrated. An extension of the present work would be to correlate measured g-to-p conversion rates with actual measurements of OH concentrations. Furthermore, this study was conducted preferentially in fair weather when free S-2

13 radical reactions would be at their maximum importance. More data should be gathered under cloudy weather conditions when aqueous processes might dominate. The question of the source of the variation in the fraction of new particle volume that appears as new particles has not been fully answered by this study. More precise data on particle surface area, volume formation rates, collision frequencies, and particle nucleation rates are necessary to determine possible relationships between these parameters and the fraction of new particle volume that forms new particles. The use of a refined version of the theory of McMurry (27) in the PHOENIX plume model appears to be a potentially useful tool in attacking this question. Further data are needed on nitrate concentrations in power plant plumes in order to evaluate more fully the importance of nitrate formation to the overall g-to-p conversion process. In view of the low nitrate concentrations that we measured, much larger sample volumes will be needed than those obtained in this study; volumes as high as 2000 liters may be necessary. Analysis for particulate organics should also be carried out together with a more detailed analysis of background hydrocarbons to identify possible particulate organic precursors. The results of this study suggest that regional differences may exist in the g-to-p conversion process. Sufficient data should be gathered at each of several different locales to allow parameters such as the ratio of particle nucleation rate to particle volume formation rate to be determined with great statistical precision at each site. Statistically meaningful intersite comparisons could then be made. S-3

14 Section BACKGROUND AND SCOPE OF STUDY BACKGROUND Although the dispersion of industrial plumes has been studied for many years, it is only recently that much attention has been paid to the complex physical and chemical processes which can take place within them. A subject of particular concern in recent years has been the conversion of SOo into sulfate particles in the plumes from coal-fired power plants. In an extensive review of this subject in 1976, Levy et a1 (1) concluded that only minimal success has been achieved in the understanding of this subject. Since then, three important studies of gas-to-particle (g-to-p) conversion in the plumes from coal power plants have been reported. In the first of these studies, Forrest and Newman (2) describe simultaneous measurements of the concentrations of SOo and particulate sulfate at various ranges downwind from the stacks of four coal-fired power plants located in the eastern United States. Both particulate sulfate and SOo were measured by the filter pack method [Newman et a1 {3}]. Forrest and Newman concluded that the rates of oxidation of SOo to particulate sulfate in the plumes were in the range of 0.5 to 5%/hr. No distinct correlations of g-to-p rate with travel time, temperature, relative humidity, time of day, or atmospheric stability were discernible from this study. Another study of g-to-p conversion in a coat power plant plume was reported by Gillani et a1 (4_) who obtained measurements on two occasions in the Labadie power plant near St. Louis. SOo oxidation rates were again deduced from simultaneous measurements of the total gaseous sulfur and particulate sulfur at the various points in the plume; gaseous sulfur was measured with a Meloy flame photometric detector and particulate sulfur (assumed to be SO" was determined by the flash volatilization technique [Husar et a1 (5j]. It was concluded that the g-to-p conversion rate was less than 3%/hr on both days, with the highest rates occurring at peak sunlight hours and the ratio of particulate to total sulfur being linearly related to the total solar radiation dose experienced by the plume. 1-1

15 Finally, Whitby et a1 (6) and Cantrell and Whitby (7) deduced g-to-p conversion rates by measuring (indirectly) the changes in the total particulate volume in the plume with travel time and attributing these changes to g-to-p conversion. Based on two sets of measurements in the Labadie power plant plume, Whitby et a1 deduced a g-to-p conversion rate of less than 0.5%/hr at 2 km downwind from the stack and 4.9%/hr at 45 km downwind. Cantrell and Whitby analyzed another three sets of measurements from the Labadie plant. Two of these measurement sets showed an increase in g-to-p conversion with travel time, the rates varying from 0.41 to.12%/hr; the third set of measurements produced a g-to-p conversion rate of about 0.45%/hr which did not increase significantly with travel time. G-to-p conversion is only one of many processes which can affect the sizes, concentrations and nature of the particles in the plume from a power plant. Primary stack emissions are also modified by the diffusion, coagulation and gravitational settling of particles. Measurements of the size spectra of particles at various distances downwind of coal power plants are even scarcer than g-to-p conversion measurements, mainly because the necessary instrumentation (particularly for particles less than a few tenths of a micrometer) has only very recently become available. The only reasonably comprehensive measurements of the size spectra of particles in power plant plumes which have been previously presented are those of Whitby et a1 (8) and Cantrell and Whitby for the Labadie Plant. SCOPE OF STUDY For the past three years we have been engaged in a detailed study of power plant plumes. In a previous EPRI report [Hegg et a1 (9)~\ and a published paper [Hegg et a1 (10)] we have described our studies of the reactions of ozone and nitrogen oxides in power plant plumes. The present report is concerned with observational and theoretical studies of the evolution of particles in the plumes of coal-fired electric power plants. We describe first the results of airborne measurements of particle size distributions and g-to-p conversion rates in the plumes from two large coal-fired power plants. We then describe a sophisticated 3-D numerical model designed to simulate the principal physical and chemical processes occurring in the plumes from coal power plants which affect the concentrations and size distributions of particles. Finally, the predictions of this model are compared with our field measurements. 1-2

16 Section 2 POWER PLANTS STUDIED AND AIRBORNE INSTRUMENTATION POWER PLANTS The two coal-fired electric power plants which have been studied are the Centralia Plant in Washington State and the Four Corners Plant in New Mexico. The Centralia power plant is situated in an east-west oriented river valley about 125 km south of Seattle in the western portion of Washington State. It is subject to the prevailing rainy weather characteristic of Western Washington and occasionally, when the winds are from the north, to fairly high ambient levels of pollution from the Seattle-Tacoma urban-industrial area. The plant burns coal with an ash content of 15%, sulfur content 0.5%, and an average heat content of 1.83 x 10 J kg"1 (7900 Btu 1b"1 ). The flue concentrations of SO and N0 are both about 500 ppm. Direct particle emissions from the stack are relatively low due to the use of two electrostatic precipitators (a Kopper and a Lodge-Cottrell in series which have a 99" "% efficiency in removing particulate mass. Consequently, the plume from the power plant stack is essentially invisible. The Centralia power plant is rated at about 1400 MM electrical output at full load; the electrical outputs during the fl ights to be discussed in this report are listed in Table 2-1 The Four Corners coal power plant is situated in a relatively arid region near Farmington, New Mexico, far from any major urban or industrial centers. The plant burns coal with an ash content of 22.5%, sulfur content of 0.7% and an average heat content of 2.09 x 107 J kg"1 (9000 Btu Ib t The flue concentrations of SOo and NO from the Four Corners Plant are similar to those at Centralia. However, the particle mass loading of the flue gas at Four Corners is considerably higher than at Central ia; this is due, in part, to the somewhat less efficient (97%) electrostatic precipitators on the two larger generator units and, presumably, to the higher ash content of the coal used at Four Corners. Most of the particulate mass in the emissions from the Four Corners Plant is in relatively small numbers (102 cm" of large particles (5-20 pm in diameter) The plume from the Four Corners Plant is generally visible. This is due both to the 2-1

17 particulate loading and to the relatively low rates of mixing of the plume with the ambient air under the stable atmospheric conditions which are common in the Four Corners area, particularly in the early hours of the morning when our measurements were made. The Four Corners power plant is rated at about 2100 MW electrical output at full load; the electrical outputs during the flights to be discussed in this report are listed in Table 2-1 Table 2-1 ELECTRICAL OUTPUTS OF THE POWER PLANTS DURING THE PERIODS OF MEASUREMENTS DISCUSSED IN THIS PAPER Electrical Output Date Power Plant of Plant (MU) October Centralia 640 November 5, March 16, March 30, April 28, October 8, October 19, September 22, October 18, 1977 H85 October 20, June 23, 1977 Four Corners 2050 June 24, June 25, June 29, AIRBORNE INSTRUMENTATION All of the measurements to be described in this report were obtained aboard the University of Washington s B-23 research aircraft. A complete ist of the instruments on this aircraft is shown in Table 2-2. Since the details on the instruments and methods of on-board sampl ing have been given by Hobbs et a1 (1_1_) only a summary wilt be given here. 2-2

18 Table 2-2 SPECIFICATIONS ON RESEARCH INSTRUMENTS ON THE UNIVERSITY OF WASHINGTON S B-23 AIRCRAFT Typ* (ind Typ (id Error) point" PrIHlf, Liquid contfitt cipc1uncf cipcclunc* bridge Crtir1dge S/it-a NctMrolooy Nuoro1ogy (;O.TC) (lo.s C) (1*C> (i0.2i) i-1 (0.21) a.2/3 i-1 (:loa) (101) -3-3 Ang) Ptatognph.. partlclil p*rttcl*t pirtldm p*rt1c1tl photo-trtc (c;) t03> oblkty ni1ytr 1<9ht-tClttr1ng 11gtit-icttr1ng llght-icituring Htloy Sylfn Royco (In-tour Kdlfled) toyco (In-hour idlfltd) Syita Huluring (0.5-) MO Ppt) (t7pi*) (TlO Ppb) Typn hydromtwri Typi (ttrtlcltl p*rt1cl pttlcle npllotor Optlcil pourlzitlon Light- Iteforologj. Ittforology I"1 (detttu pttldn u) CR-3 (noil ruclcl * Light Systaa Sntm Nuiurtng ttetur1ng c 3 (pirtlcitt L-I m2 nudl PDtirliIng tchn1qu Bollaj (Modified 1n-houn) I*1 Ducleport/ HHIiporc 3000,m.-3 (i0.1 ug-3) conulning pttlclts TI-1 frr* plotter gww*tor AM/APIIZ2 Mil-10 Syltron Corp. (H) (tsl) (1!) (1:105) chirge1" p1" proctsitnq (thm im)1<) hard-copy floppy ugnetic UghtlCtfr1ng HMdinc" Infnting nephclgactcr Gyroc HlUorologr Sp<rry (tik) (t2l) " pjrtlcit d1sp1yd study. ipnd inflit DoppiT Birrltr-ltyT Epplty Liboratory.-; l-1 (t5l) 2-3

19 In order to obtain essentially point measurements of the size spectra of the particles and the masses of participate sulfate and total aerosol in a plume, "grab samples" of plume air were sucked into large bags in the aircraft (at a rate of 100 A s~ for the particulate sulfur measurements and 4 a s~1 for the particle size spectra measurements). These samples were then passed immediately into the appropriate instruments (TSI 3030 and 3205, Royco 202, and the Pallflex filters) and the data from these instruments (except for the filters) were fed into an on-board computer. The computer generated hard-copies of the data (and various derived parameters) as well as storing them on a floppy disk for subsequent detailed analysis. "Grab samples" were generally taken in regions of a plume where the continuously measured variables (e.g. gaseous sulfur, NO 0-, 0 and Aitken nuclei indicated the plume to be most pronounced. Each plume was generally sampled at a number of different altitudes (at least three) at various distances downwind from the stack. In most cases, each measurement presented in this report was obtained from one aircraft pass through the plume. However, on all flights a few repeat passes were made through the same part of a plume in order to check reproducibil ity of the data. Measurements obtained at plume center in these repeat passes agreed to within 10%, but near the edges of a plume the variabil ity increased to 100%. The variability of the ambient air samples over short time intervals (<10 min) was less than 10%. On some flights however particle loading in the ambient air increased or decreased with time. For the g-to-p conversion estimates based on the total aerosol volumes, the "grab samples" taken at each range were averaged in the vertical (for the g-to-p estimates based on Aitken nuclei formation, plume average concentrations were also employed). The Pallflex filters, from which g-to-p conversion rates were also deduced, were exposed only to the "grab samples" taken at the altitude where the SOn reached a maximum value. On some occasions pilot balloons were released and tracked from the ground to obtain wind measurements, however, this was not essential since continuous wind measurements (as well as many other meteorological variables see Table 2-2) were measured aboard the aircraft. * Prior to being passed into the particle size measuring instruments the air samples were either heated or passed through a diffusion drier. Consequently, the particle sizes referred to in this report are those at which the particles are tn equilibrium at low (S40%) relative humidities. 2-4

20 Section 3 DATA ANALYSIS TECHNIQUES PARTICLE SPECTRA The particle size spectrum measurements are processed by a ground-based computer to give particle number (dn/dlog D) particle surface area (ds/dlog D) and particle volume (dv/dlog D) concentration distributions (N, S and V are the total particle number, surface area and volume concentrations and D the effective particle diameter). Up to three log-normal particle number concentration distributions are then fitted to the measured distributions using a -mmmzat-ion scheme [Bevington (TJ?)] which uses an algorithm given by Marquardt (1_3) the number of distributions used to fit the data is determined objectively. Each log-normal distribution describes a "mode" of the particle size distribution. Willeke and Whitby (M_) refer to these modes as "Aitken nuclei", "accumulation" and "coarse particle" modes; we will refer to them as modes 2 and 3, respectively, and indicate them, when necessary, by the corresponding subscripts. The particle size distributions generated by the fitting routine described above are used in our numerical model of power plant plumes which employs particle volume (rather than diameter) as the independent variable, thus a transformation from diameter to volume has to be made. In this case the transformed experimental data (dn/dlog v versus v) are fitted to the following function: 3 N. exp in2 v 2 Svz (3-1 where v is the volume of a particle of diameter D, N the number concentration of particles in mode i and a the geometric standard deviation of the geometric v th mean particle volume v_ of the i mode. After the data have been fitted to gi Eq. 3-1 volume parameters may be transformed back to diameter parameters. 3-1

21 For example: gnz \ v vg \Si 1/3 W " W I,3 1n \Qi "gvz P where, D, D ( and D y are the geometric mean diameters with respect to the (3-2) number, surface and volume concentrations, respectively, of mode i. The geometric standard deviation a of the geometric mean diameter is the same whether it is evaluated with respect to the number, surface or volume concentration; for the i mode it is given by: \1/3 "gdz "gvi (3-3) It should be noted that unlike Cantrell and Whitby, who make various assumptions about the distributions (e.g. they assume that the geometric standard deviation o p. of the geometric mean diameter of the "Aitken nuclei" mode is.5 +/- 0.1 we place no a priori restrictions on the modal parameters. GAS-TO-PARTICLE CONVERSION RATES Three methods are used in this paper to deduce gas-to-particle (g-to-p) conversion rates in plumes. The most direct method used for determining the rate of g-to-p due to sulfur oxidation alone was to measure the ratio of particulate sulfur to gaseous sulfur at various points in the plume. We used the flash volatilization technique described by Roberts (1 5) and Husar et a1 (5J to measure particulate sulfur. However, since we have only recently refined this method to adequate precision, only limited results from this technique are available. G-to-p conversion rates are also determined from the changes in the total volume of particles in the plume with travel time. The change with time in the volume concentration V of particles in an expanding plume, assuming an expanding 3-2

22 Lagrangian box model is given by: dv dt -3(1n V ---t - MA) + R (3-4) where V is the volume of the plume, Vn the volume concentration of particles P in the ambient air, and R (which is assumed to be constant) is the rate of production of particle volume by g-to-p conversion in a unit volume of the plume. Also, we may write: 9(1n Vp) 8t (3-5) where a is a diffusion parameter which can be determined by either the downwind change in the cross-sectional area of the plume or the change with travel time in the mean concentration of a conserved species in the plume [Hegg et a1 (9j]. Noting also that: - dt dv dt and combining Eqs. 3-4 through 3-6, we obtain; (3-6) da t where <j) a V-V,,. R (3-7) The solution of Eq. 3-7 is: Rt /, o \ t \ " ctto crrrat-/ (3-8) where (}> ()> (t,). Solving Eq. 3-8 for R we obtain: JL_ t" v n + \a + At. Y (3-9) 3-3

23 G-to-p conversion rates were calculated from Eq. 3-9 using both the particle volumes determined from the raw data and from the fitted particle volume distributions described above. The third method which we have used for calculating g-to-p conversion rates is based on measurements of the rates of production of Aitken nuclei in the plume. Based on the work of Luria et a1 (J_6) we assume that all of the Aitken nuclei produced by g-to-p conversion are \im in diameter and consist of sulfuric acid, The rate of production of Aitken nuclei N in the plume, employing the plume model of Gelinas and Walton (J_7) is given by: -K N N S (3-10) where N,, is the concentration of Aitken nuclei in the ambient air, K a coagulation coefficient, and S the rate of production of Aitken nuclei by g-to-p conversion. The solution of Eq is in the form of imaginary Bessel functions, and since S appears in the argument of the Bessel function it cannot be solved for explicitly. An asymptotic solution to Eq. 3-10, given by Hegg et a1 (9), is of the form: (1 + f) V (3-11 While S can readily be obtained from Eq the values are only applicable at large t (200 mins). For the shorter travel times appl icable to our data, NN and Eq may be written as: d N2? + ST (3-12) also 6P_ ap_ dt t where P is the concentration of a conserved species in the plume (again assuming PP for short travel times). In this study we have chosen SOo as the conservative tracer, since the amounts by which its concentration changes are negligible over the short travel times with which we are concerned, and we have assumed that the 3-4

24 rate of production of Aitken nuclei by g-to-p conversion is proportional to the concentrations of SO, so that S S P. From Eqs and 3-13: or, ik?) -o (3-14) K + S- (3-15) where N/P and we have assumed that N may be approximated by the average Aitkennucleus concentration N over the relatively short time intervals with which we are dealing. The solution to Eq is: - (o f- + K N -l -P N \ K N / L o O J uhprp where ill V ilift (t }). Frnm From Fn. Eq..?-1fi" 3-16: o o ofr P [- o s- ----L (3-17) exp [- Kp N (t-ty)] and the rate of production of Aitken nuclei is given by S ST" where P is the average concentration of SO? over the travel time considered. This method requires a knowledge of the coagulation coefficient K We assume that most of the loss in the number concentrations of particles is due to coagulation among - ( the smaller particles (mode and determine (from the particle size distribution fi ts) the val ues of F, ( ) and -o where r -? - A K is then determined from [Fuchs (18)]: y 2 kt + "o 3 r ~ [ 9 [-J. (3-18) where k is Boltzmann s constant, T the absolute temperature, n the viscosity of air, 1 the mean free path of air molecules and A is the constant which we have 3-5

25 taken to be equal to.12 [Fuchs (18)]. Clearly, this method for calculating g-to-p conversion contains several approximations and may therefore be expected to yield cruder estimates than the other two methods. It should be noted that the first method described above measures only sulfur oxidation whereas the other two methods should measure g-to-p conversion rates due to all chemical species in a plume. 3-6

26 Section 4 RESULTS OF FIELD MEASUREMENTS PARTICLE SPECTRA Shown in Figure 4-1 are the fitted particle number distributions based on measurements made at points close to the centers of plumes from the Central ia and Four Corners power plants. (The center of a plume in a horizontal traverse was taken to be the point where the SO? concentration reached a maximum value. No spatial averaging or extrapolations of the measurements have been made. This differs from the technique used by Mhitby et a1 (8) who calculated total particle volume fluxes on the assumption that a direct correlation existed between the light backscattering coefficient and particle volume. Our data show that in the case of the plume from the Centralia power plajit this assumption would not be valid, since most of the variations in the total volume of particles in the plume occur in mode and the particles in this mode are too small to cause appreciable variations in the backscattering coefficient. Complete istings of the modal parameters derived from our measurements at the Central1a and Four Corners power plants are given in Tables 4-1 and 4-2. The parameters associated with the three particle modes fall within fairly distinct ranges, as illustrated by the values shown in Table 4-3. Modes 2 and 3 sometimes appeared and disappeared with travel time in the plume: we attribute this to these two modes occasionally overlapping one another. The differences in the number concentrations of particles in mode 3 in the Four Corners and Central ia plumes are clearly evident from the results shown in Fig. 4-1 The concentrations of ym particles in the Four Corners plume were one to two orders of magnitude greater than in the Centralia plume. This appears to have been due primarily to the different stack emission rates of these large particles. At Centralia, mode 3 never contributed more than 12% to the total volume of particles and only 0.01% to the total number of particles. At Four Corners, on the other hand, mode 3 contributed up to 50% to the total volume and 0.2% to the total number of particles. 4-1

27 Figure 4-1 Fitted curves (from Eq. 3-1 and 3-2) to particle number concentration distributions for: (a) (b) (c) (d) (e) (f) Central ia Centralia Central ia Central ia Central ia Central ia plume plume plume plume plume plume on on on on on on November 5, The wind speed was 35 km hr~1. March 16, The wind speed was 39 km hr-1. March 30, The wind speed was 17 km hr~1. April 28, The wind speed was 9.3 km hr~1. October 8, The wind speed was 19 km hr"1. September 22, The wind speed was 8.0 km hr

28 D(,um) P Figure 4-1 (continued) (g) Central ia plume on October (h) Centralia plume on October (i Four Corners plume on June (j) Four Corners plume on June (k) Four Corners plume on June (1 Four Corners plume on June The wind speed was 15 km hr~ The wind speed was 4.8 km hr~ The wind speed was 11 km hr~1. The wind speed was 14 km hr The wind speed was 6.0 km hr~1 The wind speed was 14 km hr~1. 4-3

29 Table 4-1 MODAL PARAMETERS FOR PARTICLE DISTRIBUTIONS IN THE PLUMES FROM THE CENTRALIA POWER PLANT AND IN THE AMBIENT AIR. SEE TEXT FOR KEY TO SYMBOLS (a) General information for each data set, indicated by tine number Line Number Date/Local Time November 5, March 16, March 30, April 28, October 8, October 19, September 22, October 20, Plume (P) or Ambient (A) Air Altitude (m, MSL) Distance from Stack (km)

30 Table 4-1 (cont) CENTRALIA (b) Mode (for general information on each data set, indicated by a line number, see (a) above) Line Number n (Numbeir cm" -" x 107 x 107 x 106 x 104 x 104 x 104 x 105 x 10 x 105 x 104 x 106 x 104 x 104 x 104 x 10" x 103 x 103 x 103 x 105 x 105 x 105 x 105 x 104 x 104 x 103 x 104 x 103 x 105 x 104 x 104 gn1 (in units of 10-3 urn) (pm- cm-3) "gsi (in units of 10-2 ym) (pm3 v! cm-3) D, gvi (in uinits of 10-2 urn) gd

31 h Line Number (Number cm"-) cm~) gn2 (in pm) S2 (pm cm~) (in pm) (urn" cm"3) "9V2 (in \im) ad

32 Line Number (Number cm~) "9N3 (in pm) h 9 (pnr cm -) gs3 (in pm) V3 (pm cm~) D (in /3 pm) gd

33 (for general (e) Fraction information on (fy, each (cont) Table 4-1 CENTRALIA of the total particle data set, indicated by volume a line in mode i number, see (a) above) Line Number Total (pm3 V cm-3)

34 Table 4-2 MODAL PARAMETERS FOR PARTICLE DISTRIBUTIONS IN THE PLUMES FROM THE FOUR CORNERS POWER PLANT AND IN THE AMBIENT AIR. SEE TEXT FOR KEY TO SYMBOLS (a) General information for each data set, indicated by line number Line Number Date/Local Time Plume (P) or Ambient (A) Air Attitude (m. MSL) Distance from Stack (km) Travel Time from Stack (hrs) June 23, June 24, n June 25, II June 29,

35 \a auuvti; Line Numbe r (Numb er cm 3) x 103 x 103 x 104 x 103 x 105 x 105 x 104 x 105 x 103 x 107 x 10 x 10" x 103 x 103 x 103 x 103 x 103 ur 3N1 (in n ts of \im} ? (urn cm no (in 9 51 urn ts of urn) (urn- cm"") "gvi (in units o 10-2 um) f adl

36 Line Number Table 4-2 (cont) FOUR CORNERS (c) Mode 2 (for general information on each data set, indicated by a tine number, see (a) above M "2 (Number cm"3) "9N2 (in units of 10-3 pm) (ym cm"3) (in units of 10-2 pm) (ym3 cm"3) gv2 (in 10-2 urn) units of gd

37 Line Number Table 4-2 (cont) FOUR CORNERS (d) Mode 3 (for general information on each data set, indicated by a ine number, see (a ) above N 3 (Number cm 3) "a N3 (in u nits of 1Gi~3 pm) S (pm cm-3) D (in gs3 units of 0-2 pm) V "3 (pm3 cm-3) gv 3 (in un its o ") gd

38 Table 4-2 (cont) FOUR CORNERS (e) Fraction (f of the total particle volume in mode i (for general information on each data set, indicated by a tine number, see (a) above) Line Number VI Total V (pm3 cm-3)

39 Table 4-3 RANGES OF VALUES FOR THE GEOMETRIC MEAN DIAMETER D (WITH RESPECT TO VOLUME) AND ITS GEOMETRIC STANDARD DEVIATION o FOR THE THREE PARTICLE MODES Mode {i} "gv-i {vm} gdz i i 2 i

40 The existence of the third particle mode appears to have a substantial effect on the mechanics of the overall particle size distribution in power plant plumes. This can be seen from Table 4-4 which lists the rates of transfer of particle volume from mode to mode, in terms of the fraction of the total particle volume. For travel times less than about 15 min the third mode in the Centralia plume gains fractional volume, whereas modes and 2 lose volume (November 5, 1975, March 16, 1976 and October 8, 1976). Since the total particle volume also increased on the November 5 flight, all of the new particle mass must have been going into the third mode. The large decreases in the total volume of particles in the Central ia plume on March 16 and October 8, 1976, were probably due to plume diffusion close to the stack. However, since the fractions of the particle volumes in mode 3 were increasing, coagulation or g-to-p conversion via condensation into this mode was also taking place. Unfortunately, we do not have any measurements at Four Corners from which we can deduce changes in particle volumes for travel times less than 15 min. However, at Four Corners the fractional particle volume in the third mode increased for travel times between about 1/2 hr on June 23, 1977, June 24, 1977 and June 29, 1977, and on June 25, 1977 it increased between 2 and 3 hrs of travel time. The flights of June 23 and 25, 1977 at Four Corners also show that the third mode was absorbing the largest part of the total increase in particle volume. df,,~ yen was positive, a pg or N3 (or both) also increased with time; this suggests that in addition to g-to-p conversion, inter-modal particle coagulation was responsible for the increase in the volume of particles in mode 3. Since the rate of increase in the fraction of the total volume of particles in mode 3 is positive, it seems likely that the most important inter-modal coagulation rates in the plume were those due to particles in mode 3 coagulating with particles in modes 2 and However, after about 5 min travel time at Centralia, or to 3 hrs at Four Corners, the coagulation of particles with mode 3 should become negligible due to the dilution in the concentrations of particles in mode