ACOUSTIC Michael

SEEDING

Po Foster and John C. Pflaum School of Meteorology University of Oklahoma

Abstract. The "influence of acoustic energy on a cloud is examined using a numerical model. Results indicate that high-intensity sound applied for a few seconds leads to a shift in droplet spectra toward larger si~e~, which, in some cases, leads to earlier development of precipitation~ around airports by agglomeration of droplets to precipitation size. The device was to consist of a set of whistles arranged in a gimbaled frame, to be attached to an aircraft. When the aircraft flew over or through the fog, the device would emit a sonic beam which could be swept back and forth like a searchlight to saturate the cloud with acoustic energy. Boucher does not mention whether the device was actually ever assembled or tested. Significant advances in the application of sonic agglomeration came during the World War [I years when sirens manufactured by U.S. firms offered efficient high energy output sound sources. Since then, industrial applications haue progressed, e.go, for removal of particulate matter from industrial plant exhausts. Shaw and Rajendran (1979) propose that acoustic agglomeration could control the radioactive sodium particles that would result from a fire in a liquid metal fast breeder reactor plant Application of acoustic agglomeration to the atmosphere has also been attempted Most of the work, from World War II through the mid 1950s, was directed toward clearing fog from the vicinity of airfields. A series of field experiments, reviewed by Boucher (1960) showed that acoustic treatment of warm fogs increased visibility as much as i00 percent. This success seemed to be limited to quiescent conditions, and no improvements in visibility were noted when wind speeds exceeded 5 to 10 m.p.h. Nevertheless, the important point was the demonstration that acoustic energy can modify cloud droplet spectra. One hypothesis for increasing rainfall has been to stimulate the collisioncoalescence process in clouds in which it is non-existent or relatively inefficient. Traditionally, this method, categorized as hygroscopic seeding, has involved the injection of salt nuclei, salt solutions, or water droplets into a cloud specimen (Dennis, 1980) o While these efforts have produced promising results Ce.g., Braham et ale, 1957), widespread use has been

1.

INTRODUCTION

The concept that explosions or loud noises can affect the weather is interwoven throughout the history of weather modification. One dominant thread has been the association of war and rain, which apparently was first suggested over two thousand years ago by Plutarch " (Byers, 1973). Spence (1980) documents some of the theories and adventures which accompanied this train of thought, ranging from hail suppression cannons in Europe to balloon borne charges of dynamite in various locations around the United States. Meteorologists of the time generally took a dim view of these efforts because little scientific evidence supported any weather modification claims. However, as shall be seen, these "nonsensical" rainmaking efforts may not have been as far-fetched as they first appeared. In developments unrelated to these, a number of experiments were conducted in the late 1920s (e.g., Wood and Loomis, 1927) to investigate the properties of high intensity ultrasonic vibrations in liquids. The results led to similar experiments in gases where particulates suspended in the gas tended to cluster and form aggregates. Thus began research in the field of acoustic (or sonic) agglomeration (or coagulation) described by Mednikov (1965), perhaps the most notable of the early experiments were those of Brandt, Freund, and Hiedeman using tobacco smoke suspended in air (see, e.g. Brandt and Hiedeman, 1936). Among other things, they showed the strong frequency dependence of the phenomenons and discovered that agglomeration occurred more readily when sound frequencies were reduced to the audible range Through the 1930s, the interest in exploring the fundamental nature of the subject waned while the search for practical applications continued. According to Boucher (1960), interest in meteorological applications of the process first occurred in 1931 when A. Amy took out a UoS. patent for a device to clear fogs

38

prevented by the logistics of implementing the technique. The required large quantities of the hygroscopic agent, problems with caking and proper sizng, and the overall corrosiveness of the material have combined to cast hygroscopic seeding in an unfavorable light. Nevertheless, the thought of enhancing the efficiency of precipitation formarion in some types of clouds via stimulation of the collision coalescence mechanism remains attractive. This article examines the possibility that acoustic energy could be used to stimulate precipitation development. The preliminary work to be described in this article was based on the development by Chou~ Lee, and Shaw (1981.) of numerical model to simulate the acoustic agglomeration process in an aerosol mixture which gave good agreement with experimental results. The present study used a similar numerical mode], to examine the effects of acoustic agglomeration in an environment more characteristic of clouds 2o THE MODEL EQUATIONS

droplet radius, velocity of the

and v_ is the droplet.

terminal

Previous work on acoustic agglomeration has been mainly concerned with highly concentrated aerosols contained within small volumes over which it is feasible to maintain high sound inten-. sitieso Chou et alo, concerned with conditions such as those just listed, used four additional collection kernels, one for each of the following acoustic interactions: orthokinetic, hydrodynamic, diffusional turbulent, and inertial turbulent (For additional discussion on the nature of these kernels, see Mednikov (1965), Chou et al. (1981), or Foster (1984)). Practical limitations suggest that high intensity (>150 db) sound cannot maintained over the volumes encountered in work with atmospheric clouds. For this reason, the model used in this work neglects the effects of turbulent processes which become important at sound intensity levels of around 160 db. Thus, the acoustic agglomeration kernels used here are for the orthokinetic and hydrodynamic processes and are the same as those used by Chou et al. (1981). the orthokinetic collection kernel derived by Mednikov (1965), the cross sectional area of the collision volume is the same as for the gravitational kernel However, the length of the volume is twice the relative displacement amplitude of the drops multiplied by the frequency or number of oscillations per unit time (f). The collection kernel may be written KOK = 2 ~ ~ EOK 2 (al+a 2) AI2 f (3)

The model, used here is based on the stochastic collection equation (e.go~ Pruppacher and Klett, 1980) given by: ~n ~-{(x;t)

x/2
= / K(x~;xgx ;) n(x-x;,t) 0 f K(x~,x) 0 n(x~t) ~ n(x~t) ~

(i)
- n(x,t)

~ where K is a collection kernel, x and x are droplet volumes, and n is the number concentration per drop volume at time to The first integral represents the gain in number of droplets of volume x due to collisions between droplets with volumes less than that of x. The second integral represents a loss of droplets of volume x due to collisions with drops of all other volumes, resulting in drops of volume greater than x. The concept of a collection kernel is based on the volume swept out per unit time by a colliding droplet pair The volume is called a collision volume and its cross section is the geometric cross section multiplied by a collection efficiency. The length or height of the volume is the distance traveled by the approaching drops per unit time The most generally recognized collection kernel is that for gravitational sedimentation given by Kg = Eg ~ (al+a2)2 (Vtl-Vt2)

where AI2 is the relative amplitude of vibration of the two droplets, and 8 is a refill factor for small dropl~ts entering the agglomeration volume. The second acoustic kernel hydrodynamic kernel, given by KHy = EHy 2 ~ (al+a 2) I Ure is the

(4)

The physical basis of this kernel is directly analogous to that of the gravitational kernel. Here the relative motion of the two droplets (ure .) is. due to acoustic forces rather than gravity. A few comments about the collision efficiencies in the collection kernel equations are appropriate. The collision efficiencies used differ for each process. For the gravitational kernel, the collision efficiency is that used in the Leighton and Rogers (1974) version the Berry and Reinhardt (1974a) SCE model, which uses the Shafrir and Neiburger (1963) scheme for pairs with 1 > 30 u and Hocking and Jonas (1970) for pairs with a I, a 2 < 30 ~m. Following

where Eg is the collection efficiency due to gravity of a subscript 1 size droplet for a subscript 2 size droplet, a is the

Chou et al. (1981), the formula of Langmuir (1948) was used for the orthokinetic collection efficiencies and the formulation of Scott and Chen (1970) for the hydrodynamic efficiencies. 3. RESULTS

Methods for evaluating the SCE are discussed by Pruppacher and Klett (1980), including the method of Berry (1967), Berry and Reinhardt (1974 a,b), and Leighton and Rogers (1974). The range sizes from cloud droplet to precipitation is approximately 1 ~m to 1 mm. In order to keep the number of size classes of the discrete spectrum to a reasonable number, Berry devised an exponential subdivision so that the jth drop radius is given by a(j) = O exp{ ( j-l)/J}. (5)

All the results are presented graphically in plots of the mass density functions of the droplet spectra, which show how the water mass is distributed over droplets of various radii. The area under the curve between two radii represents the water mass in drops of those diameters. The mean radius of the spectrum is the radius of the droplet having mass equal to the liquid water content divided by the droplet number concentration. Characteristics of the cases discussed in the foliow~ng pages are outlined in Table i. The results for case 1 are shown in Fig. I. The initial droplet concentration for this case and case 2 was 225 -3. cm This spectrum exhibits little change due to the effects of gravity, and by 15 minutes shows no indication of the development of precipitation-sized drops. Application of 140 db sound at 300 Hz for ten seconds at the start of the run causes a shift of mass up the scale to larger radii. At the end of the 15-minute run, a definite tail has formed for droplets of precipitation size Berry and Reinhardt (1974b) showed their numerical cases that once a large drop signature began to develop, mass rapidly moved up the scale to the precipitation regime. In case 2 (Fig. 2~, the effect of gravity alone is again insufficient to develop large drops daring the 15-minute run. The application of 150 db sound at 300 Hz for five seconds at the start of the run causes a shift of mass further up the scale, as compared to case 1. After i0 minutes, a secondary maximum is forming for drops of precipitable size, and the process is ~ell under way by 15 minutes. In case 3 (Fig. 3), the liquid ~ater -~ content has been increased to 2 g m with an initial droplet concentration of 450 cm -3 and the sound parameters set at 140 db and 300 Hz for five seconds. The beginning of a precipitation tail appears after five minutes and by 15 minutes a well-formed precipitation hump is evident. 1 Cases Time (s) i0 5 5 Liquid Water Content (g m-3~ 1 1 2 Droplet Concentration (cm-3) 225 225 450

In the present work, the model code used is that of Leighton and Rogers in which J = 12/(in 2). The 130 size classes range from 3.94 ~m to 6.79 mm. Each simulated cloud has an initial droplet density function depending on droplet mass which is specified by a mean radius and variance in a gamma or Pearson Type III distribution. The values selected for the atmospheric variables for the tests = described in this work were D 765 mb, T = 283K, and L = -3. 1.0 or 2.0-g m Initial model tests using only the gravitational collection process gave good agreement with the graphical results of Berry and Reinhardt (1974b). The model code was then modified to include the two collection kernels defined by equations (3) and (4). At each step, the collection kernel used in integrating the SCE is the sum of equations (i) through (4). While a simulated sound is being applied to the cloud, each time step is one second. After the sound is turned off, the acoustic collection kernels are set equal to zero and the time step remains one second. For the rest of the computer run, the only process being modeled is coalescence due to gravitational effects. The tests discussed here were limited to 15 minutes of simulated time. Table Cloud Case Spectrum Mean Var. (~m) i0 i0 i0 1.00 i. O0 io00 Freq. (Hz) 300 300 300

Spectra

Sound Intensity (db) 140 150 140

1 2 3

4O

"REVIEWED"

A t = 0 minutes

B t = 5 minutes

z 0

1O 1O RADIUS 1(~O 10OO RADIUS

1~}O

1dO0

C t = IO minutes

D t--15 minutes

z O

1~ 100 RADIUS I~uml

1000

10 160 RADIUS I,uml

1000

Fig.

1. Case

I. Cloud

spectrum

change

with

time

for 1 g/m 3 in 225 drc~s/cm 3 exposed (dashed

to 140 db sound at 300 Hz for ten seconds at beginning of time sequence line). Solid line represents @ravitational effects alone.

41

A t = O minutes

B t =5 minutes

~r UJ

1~) RADIUS [~uml

100

1000

1 RADIUS [~Jm]

100

1000

C t =10 minutes

D t =15 minutes

10 RADIUS [~um]

100

10 0

10 RADIUS t~uml

1 0

10~0

Fig.

2.

Case

2.

Cloud

spectrum

change

with

time

for

1 g/m 3 in of time

225

dro~s/cm

exposed line).

to 150 db sound at 300 Hz for five seconds Solid line represents gravitational effects

at beginnin~ alone.

sea~ence

(dashed

42

"REVIEWED"
o 5

4 minutes t =5 minutes

3,

2-

RADIUS hum]

10 1 RADIUS DJml

1000

C t =IO minutes

D t =15minutes

3-

2-

1~) RADIUS I.um)

1~0

10~)0

1~0 RADIUS

1(~0

1000

Fig. 3. Case 3. Cloud spectrum chan~le

with time for 2 ,.j.,d~3

in 450 ~ovs/cm 3 ex~.osed sequence (~shed

to 140 db sound at 300 Hz for five seconds at beginning of time line). Solid line represents gravitational effects alone.

Although a precipitation tail begins to form by 15 minutes due to gravity alone, acoustic agglomeration has produced a significant enhancement of the precipitation process. 4. S UFLMARY Principles of acoustic agglomeration, as derived from aerosol coagulation studies, have been applied to a suspension of water droplets representative of atmospheric clouds using a numerical model. Preliminary results indicate that high-intensity sound applied for a few seconds produces a significant shift in droplet spectra toward larger sizes which, in some cases, leads to a rapid development of precipitation. Additional numerical studies are planned over a wider range of cloud and sound conditions. Implementation of some type of acoustic seeding technique would necessitate an aircraft outfitted with many high intensity speakers capable of flying at an altitude just above cloud base. The technical feasibility of such an endeavor remains to be explicitly studied but preliminary investigations suggest that the logistic and economic aspects of acoustic seeding, while challenging, are not insurmountable ACKNOWLEDGEMENTS The authors would like to thank Doug Lilly, Don MacGorman, and Conrad Ziegler for helpful conaments during the course of this work, Vicky Hall and Ginger Knight for typing the manuscript, and one of the reviewers for an excellent job of editing. REFERENCES Berry, E. X., 1967: by collection. 688-701. Cloud droplet growth J. Atmos. Sci., 24,

waves. Trans. ii01-iii0. Byers,

Faraday

Soc.,

32,

H. R., 1974: History of weather modification. Weather and Climate Modification, W. N. Hess, ed., John Wiley and Sons, New York, p. 3-44. K. H.~ Po S. Lee, and D. T. Shaw, 1981: Aerosol agglomeration in high-intensith r acoustic fields. Jo Coll. [. Sci., ~, pp. 335-353.

Chou,

Dennis,. Ao S., 1980: Weather Modification by Cloud Seeding. Academic Press, New York, 267 pp. Foster, Mo Po~ 1984: The Behavior of Cloud DroD!ets in an Acoustic Cield: A Numerical Investigation. M.S. Thesis, University of Oklahoma, 90 ppo

Hocking, L. M., and P. R. Jonas, 1970: The collision efficiency of small drops Quart. J. Roy. Meteor. Soc., 96, 722-729. Langmuir, Io, 1948: The production of rain by a chain reaction in cumulus clouds at temperatures above freezingo Jo Meteorolo, ~, 175-192. Leighton, H. G., and Droplet growth coalescence in J. Atmos. Sci., R. R. Rogers, 1974: by condensation and a strong updraft. 3-1, 271-279.

Mednikov, E. Po, 1965: Acoustic Coagulation and Precipitation of Aerosols (Transl0 from Russian by Co Vo Larrick)a Consultants Bureau, New York~ 180 pp. Pruppachert H0 R~, and J. D. Klett, 1978: Microph~sics of Clouds and Precipitation, Reidel~ Boston, 714 ppo Scott, W. T.~ and C.-Yo Chen, 1970: Approximate formulas fitted to Davis-Sartor-Shafrir-Neiburger droplet collision efficiency calculationso J. Atmos0 Scio, 2~7, 698700.

and Ro L. Reinhardt, 1974a: An analysis of cloud drop growth by collection: Part I. Double distributions. J. Atmos0 Sci., 31, 18141824. , and R. L. Reinhardt, 1974b: An analysis of cloud drop growth by collection: Part IIo Single initial distributions Jo Atmos. Sci., 3--1, 1825-1831o Boucher, R. M. G., 1960: Acoustic energy in fog dispersal techniques. Ultrasonic News, ~, 11-19. Braham, R. R, Jr., Lo Jo Battan, and H. R. Byers, 1957: IV. Artificial nucleation of cumulus clouds. Meteor. Monogr., 2(11), ppo 47-85. Brandt, O., and Eo Hiedeman, 1936: The aggregation of suspended particles in gases by sonic and supersonic

Shafrir, O., and M. Neiburger, 1963: Collision efficiencies of two spheres falling in a viscous medium. J. Geophys. Res., 68, 4141-4147. Shaw D. To, and N. Rajendran, 1979: Application of acoustic agglomerators for emergency use in liquidmetal fast breeder reactor plants Nucl. Sci. En., 70, 127-134. Clark C., 1980: The University of Nebraska coln, 181 pp. Rainmakers. Press, Lin-

Spence,

Wood,

Ro W., and A. L. Loomis, 1927: The physical and biological, effects of high-frequency sound waves of great intensity. Phil. Mag., 4, 417-436~

44