AD.-A258 21211111111

Icing Prevention by UltrasonicNuceatonof.-S-upercooled

MWater Droplets in Front ofSubsonic Aircraft DTICSELECTE

DECj. 1992

October 1992

DOT/FAA/CT-TN92/38

Docum t i n file at the Technical Center

LirrAtl antic Ci ty i nternational.Airport, N.J. 08405

NOTICE

This document is disseminated under the sponsorshipof the U. S. Department of Transportation in the interestof information exchange. The United States Governmentassumes no liability for the contents or use thereof.

The United States Government does not endorse productsor manufacturers. Trade or manufacturers' names appearherein solely because they are considered essential to theobjective of this report.

Technical Report Documentation Page

1. Repo## No. 2. Government Accession No. 3. Recipsent's Catalog No.

DOT/FAA/CT-TN92/38 14. Title end Subtitle S. Report oate

ICING PREVENTION BY ULTRASONIC NUCLEATION OF September 1992SUPERCOOLED WATER DROPLETS IN FRONT OF 6. Perfomin Organisation Cod.

SUBSONIC AIRCRAFT8. Performing Organization Report No.

7. Auto•bes)

Douglas R. Worsnop, Richard Miake-Lye, and Ze'ev Hed DOT/FAA/CT-TN92/389. Porfleming Organizatien Name and Address 10. Work Unit No. (TRAIS)

Aerodyne Research, Inc. 11. Contract or Grant No.

45 Manning Road DTFA03-91-C-00038Billerica, MA 01821 13. Type of Report and Period Covered

12. Sponsoring Agencv Memo and Address

U.S. Department of Transportation TechnicalFederal Aviation Administration NoteTechnical Center 14. Sponsoring Agency Code

Atlantic City International Airport, NJ 08405 ACD-230IS. Supplementary Notes

COTR: Carmen F. Munafo, FAA Technical Center

16. Abstract

Experiments were performed in the NASA Lewis Research Center (LeRC) IcingResearch Tunnel (IRT) to explore the possible application of a novel icingprevention technique. This technique is based on nucleation icecrystallization in supercooled cloud droplets upstream of the impingementsurface and relying on evaporative cooling to remove the latent heat of fusionfrom the droplet so that it can completely freeze before striking the solidsurface. Ice accretion on the impingement surface was measured to quantifythe changes due to induced nucleation.

A flow duct was designed to provide a uniform, parallel flow from thenucleation excitation volume to the test surface, a simple symmetric airfoil.The distance between the excitation volume and the airfoil in the flow ductcould be varied, but the maximum length of over 2 meters (m) was chosen tomaximize the time for freezing. Ice accretion was measured over a range oftunnel parameters, including tunnel velocities of 50-100 miles per hour (mph)and temperatures from 0 to -36 oF. At colder temperatures, spontaneousfreezing of supercooled droplets led to reductions in observed ice accretion.However, the range of active nucleation parameters explored in this set ofexperiments had no measurable effect on ice accretion on the subject airfoil.Suggested future work includes expanding the measurement parameter space and,obtaining static measurements in a cloud chamber.

17. Key Werds 1. Distribution Statement

Aircraft Icing Document is on file at the TeehnicalIce Protection Center Library, Atlantic CityIce Nucleation International Airport, New Jersey 08405Ice Crystallization

1|. Security Cleass;. (of this report) 20. Security Cleesif. (*I this page) J 21" No"of Pages 22. PrCe

Unclassified Unclassified 57

Form DOT F 1700.7 (0-72) Reproduction of completed page authorized

PREFACE

This project is an outgrowth of a previous FAA sponsored SBIR contract (DTRS-

57-90-00132). Continued research was feasible due to the additional FAA

funding and the support of NASA LeRC Icing Research Tunnel (IRT) personnel.

Beyond making the wind tunnel available, their logistical and system support

was crucial for the experimental testing described in this report.

Ace.41gm bro

ON Y'

LL

D / A

iCi

TABLE OF CONTENTS

Section Page

EXECUTIVE SUMMARY ix

1. TECHNICAL BACKGROUND I

1.1 Introduction 11.2 Aircraft Deicing 11.3 Ice Nucleation 31.4 Ultrasonically Induced Nucleation 41.5 Droplet Freezing 61.6 Research Plan 9

2. APPARATUS 11

2.1 Flow duct 112.2 Airfoil 142.3 Transducer Assembly 142.4 Control Electronics 182.5 Transducer Power Supplies 192.6 Experimental Approach 21

3. ICE ACCRETION RESULTS 25

3.1 Research Chronology 253.2 Ice Accretion Measurements 273.3 Summary 33

4. DISCUSSION 38

4.1 Ice Accretion Observations 384.2 Acoustic Power Levels 394.3 Operating Parameters 404.4 Summary 41

5. CONCLUSIONS 43

6. REFERENCES 44

APPENDICES

A - Method of Crystallizing Supercooled Water Droplets

B - Estimate of Power Requirements for Ultrasonically InducedIce Nucleation

v

LIST OF ILLUSTRATIONS

Fixure Page

I Schematic of Ultrasonic Ice Nucleation Test 2

2 Schematic Transducer/Flow Duct Apparatus in the IRT 10

3 Rough Design Sketch of the Flow Duct 12

4 Flow Duct Mounted in IRT Test Section 13

5 Airfoil Mounted to Flow Duct 15

6 Rough Design Sketch of the Transducer Mounting Plate 16

7 The Transducer Assembly in Place in the Flow Duct 17

8 Frequency Response Curves for the PiezoelectricTransducers 20

9 Ice Accretion on 3.375 Inch Airfoil 23

10 Ice build-Up in Interior of Flow Duct 24

11 Experimental Test Plan Flow Chart 26

12 Ice Accretion Profiles for Data Set 1 28

13 Ice Accretion Profiles for Data Set 2 29

14 Ice Accretion Profiles for Data Sets 3 and 4 31

15 Average Ice Accretion Thicknesses for the Central 35Airfoil Span

16 Normalized Ice Accretion Thicknesses 36

vi

LIST OF TABLES

Table Page

1 Estimated Freezing Times and Flight Distances for 920 jm Diameter Droplet

2 Summary of IRT Run Parameters 30

3 Acoustic Power Levels 40

vii

EXECUTIVE SUMMARY

Experiments were performed to explore the possible application of a novelicing prevention technique in NASA Lewis Research Center's (LeRC) icingresearch flow facility, the Icing Research Tunnel (IRT). This technique isbased on nucleating ice crystallization in the supercooled cloud dropletsusing ultrasonic acoustic radiation. Acoustic energy in the range of 10 tt650 kilohertz (kHz) was produced with a combination of piezoelectric ceramicsingle-frequency transducers and broad-band high performance loudspeakers(tweeters with extended range to 40 kHz).

The ultrasonic energy excites a volume in the IRT droplet-laden airflowupstream of a subject impingement surface, a streamlined-tubing airfoil, toperturb liquid droplets and initiate the freezing nucleation process. Betweenthe excitation volume and impingement on the subject airfoil, evaporativecooling was expected to remove the latent heat of fusion from the droplet sothat it could freeze before striking the solid surface. The distanceseparating these two points was chosen based on estimates of the heat removalrates. Ice accretion on the impingement surface was measured to quantify thechanges due to induced nucleation.

A flow duct was designed at Aerodyne Research, Inc. (ARI) and built by NASALeRC to provide a uniform, parallel flow from the nucleation excitation volumeto the test surface. A 30.5- by 61.0-centimeter (cm) (I- by 2-foot) duct, 244cm (8 feet) long, was aligned with the flow near the center of the IRT testsection. The duct samples the most uniform region of the IRT flow anddirects it toward the subject airfoil, preventing mixing or residual flowangularity from perturbing the flow path. Thus the flow volume, excited bythe ultrasonic radiation upstream of the airfoil in the flow duct, isconvected along the duct (with minimal mixing at its edges) before impingingon the airfoil.

The distance in the flow duct between the excitation volume and the airfoilcould be varied by attaching the transducer mounting plate at one of threepositions along the duct's wall. The maximum length of over 2 meters waschosen in the reported experiments to maximize the time for freezing. The IRTvelocity was kept near the minimum, 22 meters per second (50 miles per hour),at which reliable droplet sizes and flow and droplet uniformity could bemaintained. For a given transducer/airfoil separation, the lowest flowvelocity maximizes the freezing time for the excited droplets. Experimentswere performed with two different droplet size distributions and liquid watercontent in the flow. Tunnel temperatures ranging from -18 °C to -39 °C (0 OFto -36 OF) were used with a variety of ultrasonic excitation parameters foreach temperature.

Ice accretion was measured along the 61-cm span of the airfoil over 10 minutesof ice accumulation time in the tunnel. At cold, <-34 °C (<-30 OF),temperatures, reduced ice accretion was consistent with homogeneous freezing(and non-accretion) of ice particles. However, for the range of excitationparameters explored in this set of experiments, no significant effects on iceaccretion on the subject airfoil due to ultrasonic excitation were observed.Future work is suggested for expanding the measurement parameter space inorder to more fully understand the previously reported anecdotal evidence for

ix

such acoustically-induced freezing. The proposed extensions to the currentwork include increasing the ultrasonic power densities beyond the high levels(125 decibel (dB)) used in the reported experiments and possibly performingnon-flowing measurements in a static, supercooled cloud chamber.

x

1. TECHNICAL BACKGROUND

1.1 Introduction

The ultimate goal of this research program is to determine thefeasibility of a new aircraft anti-icing technique, based onprojection of ultrasonic acoustic energy in front of aircraft.Acoustic perturbation would induce freezing of supercooled clouddroplets in front of the aircraft so that frozen ice particleswill bounce rather than accrete on aircraft surfaces. Thespecific goal of this research reported here was to demonstrateultrasonically-induced nucleation in front of an airfoil surface.

This work was preceded by a laboratory study of ultrasonicfreezing of droplets supercooled in a low pressure vacuum system.Based on that work, which was unsuccessful, ARI proposed to testthe concept at realistic, atmospheric pressure, using a suitablyequipped wind tunnel to simulate icing conditions. A schematicof the proposed test is shown in figure 1.

The tests were carried out in the NASA LeRC Icing Research Tunnel(IRT). Building on experience gained from previous FAA SmallBusiness Innovative Research (SBIR) support, Aerodyne personneldesigned and constructed an assembly containing a total of nineacoustic transducers and associated drive electronics. Inconjunction with NASA LeRC personnel, a test flow duct wasassembled in the IRT and interfaced with the transducer assembly.

During the period of 15-19 December 1991, supercooled dropletsgenerated in the IRT were irradiated with ultrasonic energy overa range of conditions. Droplet freezing was monitored bymeasuring ice accretion on an airfoil mounted at the end of thetest duct. The background and implications of the failure toreduce observed ice accretion are the focus of this report.

In the remainder of Section 1, the technical background of icenucleation in general and of ultrasonically induced nucleation inparticular are given. This is followed by a discussion of thetime required for droplet freezing. These considerations formedthe basis for the test design and research plan.

In Section 2, the test apparatus and procedure are described.Sc-tion 3 reports the results of actual test measurements.Finally, in Section 4, the results and their implications arediscussed. Section 5 gives conclusions.

1.2 Aircraft Deicing

Currently, ice protection for general aviation aircraft is pro-vided either by thermal means (heating elements or hot mediumcirculation), by mechanical means (inflatable rubber boots), orsometimes by transpiration bleeding of melting point suppressants

S- -• m m m | | I

Super CooledDroplet

Flow

Icing Shield

UltrasonicTransducer

Variable (0.1 to 2 meter)

Ice Accretion Monitor(Airfoil Surface?)

ONm

2 meters

M91-2491D.W.

Figure 1. SCHEMATIC OF ULTRASONIC ICE NUCLEATION TESTUltrasonic transducers irradiate transversely across windtunnel flow of supercooled droplets. Ice accretion ismonitored on a surface placed downstream of the transducers.Measurement of accretion thickness as a function of distancefrom the transducer will map out kinetics of thenucleation/freezing processes.

on the leading edges (glycol based deicing systems). Thesesystems are either bulky, have finite flight time capabilities orconsume large amounts of energy. A new anti-icing strategyinvolving ice nucleation and freezing of the supercooled waterdroplets prior to their impingement and solidification on theleading edges of the wings and fuselage of the aircraft wasproposed. Application of this approach would induce cavitationnucleation or perturbation nucleation in the supercooled dropletsby ultrasound emitted from a series of transducers implanted inthe leading edges or the fuselage.

If the ultrasonic energy can be delivered efficiently, then theapproach will broaden the operational capabilities of general

2

aviation aircraft by decreasing the weight and the energyconsumption relative to current systems. The goal of this projectwas to determine the basic physical parameters underlying theultrasonic nucleation process in order to assess the potential foractual implementation of ultrasonic anti-ic-ing.

1.3 Ice Nucleation

The process by which a liquid crystallizes when cooled below itsmelting point involves two distinct mechanisms, crystal nucleationand crystal growth. The driving force for freezing is thedifference in free energy between the liquid and solid phases at agiven temperature (primarily determined by the heat of fusion).

The condition for the triggering of the phase transition bWtween theliquid and the solid phases is the formation of embryonic nuclei inthe liquid. This requires the establishment of surraces between theliquid and the newly formed nuclei, which requires energy to perturbthe liquid. Since the surface energy of each embryonic nucleusincreases with the square of its radius (surface area) and thedifference in free energy between its solid state and its liquidstate increases with the cube of its radius (volume), one can seethat very small nuclei are not stable since the surfa-e energy ofthe nucleus is larger than the energy gained from fusing its volume.

Once a critical size of such an embryonic nucleus has been reached,however, the energy gained from the heat of fusion is larger thanthe energy required to increase the nuclei's' surface, and thus theirfurther growth lowers the free energy of the system. Such nucleiare therefore thermodynamically favored; they are stable andcontinue to grow until the liquid phase is converted to thecrystalline phase. The minimal radius of a stable embryonic nucleusis termed the critical radius, and is the radius at which the gainin free energy between the two states equals the energy required toform the new surface.

Differences between homogeneous and heterogeneous nucleation insolidifying liquids are. noted. Homogeneous nucleation rates dependonly on the thermodynamic properties of the liquid in question.Heterogeneous nucleation occurs on pre-existing surfaces (usuallyundissolved impurities or external solid surfaces). The presence ofsuch surfaces removes the need to supply the energy for forming theembryonic nucleus' surface, or, at least, dramatically r?•q1ces thesurface formation energy, and thus the critical radius.

For any liquid, in principle, a homogeneous solidificationtemperature can be defined, which would be the lowest possible(equilibrium) temperature at which super-cooled liquid can exist.

Due to the impossibility of maintaining systems absolutely free ofany impurities, there are always some uncertainties in thedetermination of this temperature. For water, however, atemperature of -40OF is well accepted as the homogeneous freezing

3

temperature. 1 Freezing of water at temperatures higher than thehomogeneous freezing temperature requires the presence of pre-existing nucleating surfaces.

Icing conditions in the atmospher6 are created when water dropletscool below 32°F in the absence of any nucleating agents. When anaircraft enters a cloud formation containing supercooled waterdroplets, the solid surfaces of the leading edge serve as nucleationsites for the supercooled droplets, and icing occurs. While this isa simplistic explanation of a more complex phenomenon, it is wellknown that icing does not occur at very low temperatures where thedroplets have undergone homogeneous nucleation and only iceparticles are present. It is surmised, therefore, that ifnucleation and freezing in advance of their impingement on theaircraft can be effected, then the ice particles will bounce fromthe aircraft structure and ice accretion will be avoided.

In the atmosphere super-cooled droplets often (if not usually)freeze heterogeneously on solid impurities (e.g. dust or sootparticles) within the droplet. Heterogeneous nucleation ofsupercooled droplets has been induced with silver iodide crystals,and the heterogeneous nucleation temperature can be raised as highas -7°F. 2 An alternative nucleating agent is the naturally occurringmicro-organism pseudomonas syringae, capable of raising theheterogeneous nucleating temp;rature to the range of -5 to -90F. 3 Itis not practical, however, to use these techniques for the purposeof avoiding icing in flight. Delivery of nuclei in front of amoving plane would be cumbersome, the weight penalty too large and,moreover, the environmental impact of such an approach could beproblematic.

1.4 Ultrasonically Induced Nucleation

Therefore, the investigation of an alternative mode of heterogeneousnucleation was proposed, whereby the lowering of the surface energyis provided by ultrasonic induced density fluctuations within thedroplets (or on their surfaces). The technique, also known ascavitation nucleation, has been used successfully in thesolidification of larger masses. There is anecdotal evidence ofsuch effects for aerosols, though there has not been any systematicstudy of the phenomenon. For example, major rain discharges fromclouds with laden supercooled droplets are known sometimes to occurimmediately following a lightning stroke (and its accompanyingthunder). One can speculate that sudden acoustic pressure wavescould cause nucleation and thus agglomeration of supercooleddroplets. While the same energy density in a system of airbornetransducers is impractical, it could be possible that fine tuningwould result in a similar localized effect at much lower energydensities.

4

Reports of more direct anecdotal observations appeared in 1941.Ronald L. Ives 4 reported that, under some circumstances, noisedisturbances (mostly in the audible range) of high intensity andwith a broad spectral distribution, were capable of enhancingdroplet nucleation and coalescence in "River Fogs" (supercooledmountain cap clouds). However, the mechanisms for the observedbehavior were never elucidated and the reproducibility of theexperiments was ambiguous. Nevertheless, some indication of a non-vanishing interaction between acoustic energy and supercooleddroplets was evident.

The Ives report prompted a comment by Swinbank,5 who reported thatduring WW II that the British Meteorological Office had testedultrasonic anti-icing in a wind tunnel, with negative results. Infact, they observed enhanced ice accretion, which they interpretedas evidence of partial freezing of the supercooled droplets. Thisis actually evidence of a positive ultrasonic nucleation effect,since, as discussed below, ice crystal nucleation and liquid dropletfreezing are physically separate processes. No details of the windtunnel testing were reported; thus, it is not possible to quantifythe significance of their observations. Thus, this observation mustbe considered anecdotal and its explanation speculative, thoughphysically reasonable, i.e. partially frozen droplets may indeedhave enhance accretion efficiencies.

During the course of this project, a patent issued in 1949 for'Method of Crystallizing Super-Cooled Water Dropletsm 6 wasdiscovered. Originally directed toward fog reduction at airports,the authors did note the potential for aircraft anti-icingapplications. The patent referred to laboratory observations ofsiren-induced freezing of supercooled droplets in a cloud chamber.A copy of the patent is included as Appendix A. The patent reportsthat audible and ultrasonic frequencies up to 30 kHz were effectivein causing nucleation and that power levels of 150 dB were required.The principal author, Florence van Straten of Bethesda, MD, wascontacted and she confirmed the experimental results, but indicatedthat the laboratory results were never formally reported. Again thedetails reported in the patent are insufficient to evaluatepotential application to aircraft anti-icing. In particular, thereis no indication of the origin of the 150 dB nucleation thresholdnor of the frequency range of the acoustic output of the siren.

The van Straten patent points out the two key parameters that needto be evaluated for understanding of ultrasonically induced icenucleation: acoustic frequency range and minimum energy densi-y.Both are key to specifying electrical power and mechanicaltransducer requirements for eventual airborne anti-icingapplication. This project was designed to provide the firstsystematic study of ice nucleation effects induced by a wide rangeof acoustic frequencies.

5

Based on these anecdotal observations, the test plan includedutilization of acoustic frequencies ranging from the audible to theultrasonic. Transducers in the audible (and near audible) range arereadily available as high performance audio 'tweeter' components,which give continuous frequency coverage in the range 1 to 40 KHz.Ultrasonic frequencies (up to 1 MHz) require specification ofpiezoelectric transducers at fixed, resonant frequencies. Threesuch transducers were specified at 80, 350 and 650 KHz. Powerlevels were maximized within the limits of available transducers andtheir associated electrical amplifier/drivers.

In the proposal initiating this study, power requirements derivedfzom first principles were estimated. This discussion is includedas Appendix B. The estimate involved comparing the energetics offormation of requisite embryonic ice nuclei with available acousticenergy densities. The latter involves the intrinsic problem oflarge mismatches in acoustic impedances in thetransducer/air/droplet system. These mismatches occur between thetransmitting medium (air) on one side and the transducer's outersurface and the water droplets on the sending and receiving sides,respectively. The estimated minimum power requirements are notinconsistent with the 150 dB levels reported in the van Stratenpatent. However, because of assumptions required for thesecalculations, there is considerable uncertainty in their accuracyand applicability, hence the need for empirical measurements.

These calculations emphasize the need for minimizing the mismatch intransducer/air acoustic impedance. Building on previous FAAsupported work, 7 ultrasonic transducers with 35-40 dB losses at theair interface (as compared to 80 dB losses for conventional designs)were specified and procured. As reported in Section 2, thecapability to generate audible and ultrasonic acoustic waves atpower levels of up to -125 dB resulted.

The research plan was to test ultrasonically induced ice nucleationby irradiating a wind tunnel flow of supercooled water droplets withthe acoustic transducers described above. Such nucleation would bemonitored by measuring ice accretion on an airfoil surfacedownstream of the transducers. However, the relationship betweenice particle nucleation and sabsequent accretion of ice on a surfacedepends on the freezing of the liquid droplet in the interveningtime.

1.5 Droplet Freezing

Freezing of the entire volume of a super-cooled water droplet is aphysically separate process from the initial nucleation of a singleice crystal within the droplet. The nucleation process, asdiscussed above, is essentially a molecular process, involving theformation of small embryonic nuclei whose volume is very smallrelative to that of a cloud droplet. Once nucleation has beeninitiated, the freezing of the volume of the entire droplet must be

6

considered separately. In the context of anti-icing in flight, thequestion is: How far in the front of the aircraft must icenucleation be initiated?

The isate of freezing needs to be compared to aircraft speed and thedistance in front of leading edge surfaces. As a reference point,consider a distance of one meter which, at 400 mph, corresponds to -

5 x 10-3 seconds (s). This is very long compared to most molecularprocesses, i.e. there is no fundamental, physical process limitingwater droplet freezing on this millisecond scale. In fact, inprevious work at Aerodyne it has been shown that super-cooleddroplets can indeed freeze into "snow" in less than 10-3 S.8

On a millisecond time scale the rate of freezing is actually a heattransfer problem. Upon ice nucleation, liquid freezes until releaseof latent heat of fusion warms the droplet to the freezing point,320 F. At that point, more liquid can freeze only as heat of fusion(80 cal/g) is removed to maintain the temperature at 320 F. By farthe most efficient heat removal process is evaporation of watermolecules, since the heat of vaporization of water (600 cal/g) is somuch larger than the heat of fusion. In previous laboratory work atAerodyne, 8 molecular collision limited evaporation at low pressure(<3 torr) allowed very fast freezing (< 10-' s).

At atmospheric pressure, evaporative cooling will be limited by gas-phase diffusion of ,.ter vapor away from the droplet. The drivingforce for water vapor evaporation is proportional to the differencebetween the liquid water vapor pressure at 32 0F (e.g. 4.5 torr) andthe ambient water vapor pressure (at the super-cooled droplettemperature). The rate of gas diffusion of water vapor away fromthe droplet is inversely dependent on the droplet volume and therate of cooling for the entire droplet is directly dependent on thedroplet surface area.

The overall rate of the above processes is estimated below. Thediffusion limited evaporation rate is given by 9

An Yg c 2

RE = (mole/s)4

where the rate of diffusion is expressed as the collisionprobability coefficient

8 Dg

7g -cd

7

where An = n32oF - nT subscript denotes temperaturen = vapor density (mole/cm3 )c = mean molecular speed (cm/s)

Dg= gas diffusion coefficient (cm2 /s)d = droplet diameter (cm)

The rate of heat removal from the droplet is given by the product ofRE and the heat of vaporization of water Cv

18g/moleRH = RE Cv (cal/s)

6.1023 mol/mole

The overall freezing time can be estimated by dividing the heat offusion to be removed by this heat removal rate,

CF xd 3 (1 g/cm3 ) ( 1 - AT Cp / CFtF =

6 RH

where CF a heat of fusion (cal/g)Cp m specific heat of water (1 cal/g/°C)AT a difference in ambient temperature and

water freezing point (O°C)

The right hand term in the numerator takes into account the heatcapacity of water gained in warming the droplet to 32°F (O°C) afterice nucleation. Combining these equations, one obtains

CF d 2 ( 1 - AT Cp / CFtF =

Cv An Dg 4.10-22

This expression indicates that freezing time is proportional to thesquare of droplet diameter. The time also decreases withtemperature as both AT and An increase. Table 1 lists freezingtimes calculated for 20 pm diameter droplets at 28, 20, 0 and -30°Ftemperatures (with Dg = 0.4 cm2 /s for water in air at 400 torr).

One should note that these estimates assume there is one nucleationsite per droplet; large droplets may have multiple sites which couldreduce estimated freezing times somewhat.

The tF values in Table 1 represent the times required to freeze theentire liquid volume of a supercooled droplet. In the context ofanti-icing, these times and their corresponding distances at flighttypical speeds represent upper limits for ice nucleation underflight conditions. The key question is: How much of a liquiddroplet needs to freeze for the particle to bounce rather thanaccrete.

8

Table 1. Estimated Freezing Times and Flight Distances for20 pm Diameter Droplet

Temperature Water Vapor Flight DistancePressure An tF 50 mph 40r' mph

OF OC (torr) (mol/cm3) (sec) (meter)

32 0 4.6 028 -2 4.0 2 x 1016 0.16 3.5 2820 -7 2.7 6 x 1016 0.06 1.2 10

0 -18 1.2 10 x 1016 0.03 0.6 5-30 -34 0.2 13 x 1016 0.015 0.3 2.5

1.6 Research Plan

The work plan of this project was based on the cooperative agreementof FAA and NASA personnel that made available the NASA LeRC IRT forice accretion tests of ultrasonically induced ice nucleation. Basedon discussions with NASA LeRC staff, it was decided to coupleacoustic transducers with a flow duct mounted in the center of theIRT. The following general tasks were defined:

(1) Design and construct an assembly containing a total of nineacoustic transducers (and associated drive electronics) tocover the frequency range 10 to 650 KHz at power levels up to120-130 dB.

(2) Design and construct a flow duct (8 x 2 x 1 feet) to be mountedin the center of IRT. Mount the transducer assembly a variabledistance upstream of an airfoil. Distances (2 to 8 feet) werechosen to match expected freezing times as listed in Table 1.

(3) Monitor ice accretion on the airfoil as a function oftransducer frequency and position. In order to maximize bothice nucleation and liquid freezing processes, start withcoldest .(-30 0F) temperature, smallest (14 pm) droplet size andslowest (50 mph) air flow speed. Depending on results, varyexperimental conditions to determine envelope of icingconditions.

Figure 2 shows a schematic of the apparatus as assembled and coupledwith the NASA IRT facility.

The following three sections describe the apparatus, results andconclusions, respectively, of the testing measurements made in theNASA LeRC IRT during 17-19 December 1991.

9

CCC

U)0-

0t 0

0.U

U)U

~~E-4I-5 0

11.50

a..

LU U

00

01

2. APPARATUS

The experimental apparatus for these measurements is composed of aflow duct, an ultrasonic transducer array, power supplies andcontrolling electronics for the transducers, and computer control ofthe transducers and data logging. This apparatus was designed to beintegrated with the NASA Icing Research Tunnel (IRT) facilities andsignificant design and fabrication support was provided by the IRTstaff. The flow duct and transducer mounting plate were designed byARI with significant input from IRT staff and both were built atNASA LeRC. The transducers and their control electrenics wereassembled and integrated with the NASA fabricated components and theentire assembly was mounted and tested in the IRT in the first dayof the experimental program.

Beyond supplying the tunnel, test section, and flow with calibratedsuper-cooled droplet properties, NASA provided software forcaptioning the experimental results. Output from the operatorconsole central tunnel control system ('WDPF')l° was provided to logthe tunnel conditions and the droplet spray system operatingparameters. The transducer temperatures were monitored withthermocouples that were connected to the NASA LeRC data acquisitionsystem ('ESCORT'). 1° Print-outs of these temperatures provided alog of reference temperatures for each experimental condition.

2.1 Flow Duct

The flow duct is a 1 ft by 2 ft rectangular duct with three aluminumplate walls that is 8 ft in length, open at both ends (seeFigs. 3,4). The fourth side of the duct is a 2 ft wall made oftransparent lucite so that ice accretion on the walls andtransducers can be monitored visually while the experiments arebeing performed. The purpose of this duct is to sample a clean flowwith uniform velocity and water droplet properties near the centerof the IRT test section and direct it toward the test airfoillocated just beyond the open downstream end of the flow duct. Theduct walls prevent any large scale residual swirl or cross flow fromdisplacing and/or mixing the fluid that enters the duct before itimpinges on the test airfoil. Thus, after the moisture-ladenflowing air is exposed to the ultrasonic perturbations from thetransducers , it will convect with minimal disturbances and mixinguntil it reaches the test airfoil.

The duct mounts on legs that bring the duct inlet into the uniformcentral region of the flow in the IRT test section (see Fig. 15 inreference 10). The power leads, thermocouple wires, and purge gasline are routed from the downstream end of the transducer cover,down one support leg, and through an opening in the tunnel floor,from which they travel into the IRT control room. There, thethermocouple wires connect to the 'ESCORT' data acquisition system,the single- frequency transducer power cables connect to the ENI high-frequency power supply, and the broad-band transducers (tweeters)

'I 0

U) u

z E)

4 0

)4 -

0L -

10

Q)O

ý-lU)0

IE-4

uowro U)

110 4-)

C-43 (a

oun

102

Figure 4. FLOW DUCT MOUNTED IN IRT TEST SECTIONThe 8 foot long flow duct is mounted near the center of the 6' by 9'cross section of the 20' deep test section. The transparent face of theduct is on the right in this photograph, taken looking downstream, and isdirected toward the control room viewing windows which are visible atthe right edge of the photo.

13

are connected to an extended-range audio amplifier which powerst hem.

2.2 Airfoil

The test airfoil is mounted on plates (see Fig. 5) that attach tothe 30 cm duct walls and position the airfoil in the center of ductcross section along its long (61 cm) transverse axis. The leadingedge of the airfoil is displaced from the trailing edge of the ductso that ice accretion can be observed visually during a test withoutinterference from duct walls. Three different airfoils wereassembled, with chords ranging from 3.0 cm to 8.6 cm. The 8.6 cmairfoil was used in the measurements reported. Reference linesalong this airfoil's span allow the span-wise variation in iceaccretion to be quantified. The airfoil assemblies use streamlinedsteel tubing welded to steel endplates to form a rigid structurethat aligns the symmetric airfoil in the flow direction.

2.3 Transducer Assembly

The transducers are mounted in an assembly that mounts flush to thealuminum 30 cm wall of the flow duct (opposite the lucite wall).The transducer assembly can be mounted in any of the three locationsalong that wall, allowing the distance (and thus the time, for agiven flow velocity) between the transducer excitation and theairfoil to be selected. Shorter distances and times can be used toexplore the temporal dependence of droplet freezing, with thelongest distance chosen as the initial test condition as used inthese experiments.

The transducer assembly (Figs. 6,7) accepts an array oftransducers of four different frequency characteristics andmultiple transducers of each type were used in these experiments.The higher ultrasonic frequencies were generated by pairs of threedifferent, single-frequency transducers operating at, nominally,80 kHz, 350 kHz, and 650 kHz. Each of these pairs were arrangedso as to illuminate a streamtube that flows to the test airfoil.One transducer is displaced a little above and the other isdisplaced equally below the dividing streamline so that the entirestreamtube impinging on the airfoil is illuminated. Thisstaggered arrangement of the transducers ensures that even with areasonable amount of turbulent mixing in the duct (which has beenrelegated to smaller turbulent length scales by the confiningeffects of the duct), the fluid impinging on the airfoil has beenilluminated with the excitation energy.

The three pairs of staggered single-frequency transducers arefollowed by a set of three high-frequency audio loudspeakers(Dynaudio D-21F tweeters) that are capable of generatingultrasonic excitation up to 40 kHz. Unlike the single-frequencytransducers, these tweeters are broad-band and very uniform in

14

Figure 5. AIRFOIL MOUNTED TO FLOW DUCTThe 3.375 inch chord airfoil, composed of streamlined tubingwelded to two mounting end plates, is shown attached to thedownstream (exit) end of the flow duct. The transducer mountingplate is attached to the aluminum wall on the righthand side of thisphotograph, taken upstream of the test section. The transducers areat the upstream end of the flow duct and are thus out of view in thisphoto.

15

t~14

12

IL 07vi -~;A

z'

'p ~ V (- ~o-0

0

ONO r E-4a

ro u

tgo

ft-3

-~- -l

.I

0' -~cJ216

Figure 7. THE TRANSDUCER ASSEMBLY IN PLACE IN THE FLOW DUCTThe six single-frequency transducers, two each of 650 kHz, 350 kHz, and 80kHz from left to right in the photograph, are positioned in overlapping pairs toensure good excitation coverage of the central flow down the duct. The broadbandtweeters (20 kHz - 40 kHz) are barely visible to the right of the 80 kHz transducers.The dry purge gas slot is visible just to the left of the leftmost 650 kHz transducer.

17

their response. Their major limitation is that the total power theycan deliver is the sum or integral across the input frequencybandwidth, meaning that the power available at any given frequencyis diminished by the power simultaneously delivered at all otherfrequencies . The three tweeters are arranged with one on the ductcenterline and one on each side, again to minimize the effects ofunexcited fluid mixing into the impinging streamtube from the sidesdue to turbulent fluctuations.

The transducers and their impedance-matching components (see below)are set in the mounting plate with the emitting faces of thetransducers aligned to the flow-exposed surface of the plate. Thecomponents and connections on the other side of the plate areprotected from the droplet-laden IRT flow by a sheet metal coverthat is sealed to the back of the mounting plate. The transducerpower connections are made through bulkhead BNC fittings in thedownstream cover wall, while the thermocouple connections to thetransducer temperature sensors run through an opening in this wall,which is sealed during mounting. A Polyflo bulkhead feedthrough isused to provide dry purge gas for the transducers.

The dry-gas purge is used for two purposes. The first is to preventmoisture from accumulating on or around the electrical connectionsto the transducers. To do this, the entire internal volumecontained inside the cover is purged with the dry-gas flow. Thus,even though the cover is sealed, any leaks and resultantcondensation of existing water vapor will be swept out of the cover.The purge flow leaves the volume behind the mounting plate through aslot in the front of the mounting plate itself, venting into theflow duct at the upstream end of the transducer array (see Fig. 7).This provides a second use of the dry purge gas.

By injecting dry purge gas into the boundary layer along thetransducer mounting plate, the transducer faces are afforded someprotection from ice accretion due to th, droplet-laden IRT flow.After performing the experiments, it became apparent that thisprotection was minimal, which may have as much to do with thedynamics of ice accretion as with turbulent mixing through theboundary layer. In any case, for dry-gas injection levels that didnot significantly perturb the duct flow, ice accretion was notprevented on the transducer faces and the wall downstream of theinjection slot. The accretion that was observed was not large andcoverage was spotty, growing as whiskers on this surface parallel tothe flow, and was cleaned after each run.

2.4 Control Electronics

The ENI high-frequency power supply used to drive the single-frequency transducers has a 50 Ohm output impedance. Thetransducers, on the other hand, have a complex impedance that is arapidly varying function of frequency near the respective resonantdesign frequencies (Figs. 8 a,b,c). To optimally excite t'Ae flow

18

with the maximum power, the transducers were impedance matched tothe output amplifier of the ENI at their particular frequency byincorporating a (remotely located) high frequency transformer and anappropriately chosen inductor to offset the transducer's capacitanceat that frequency. A high value bleed resistor was also wired inparallel to the transducer since these devices also demonstratepyroelectric behavior and develop significant capacitive charge whencycled in temperature, particularly for the extreme temperaturevariations in the IRT. This small electronics package is mounted onthe back of each single-frequency transducer along with a BNCconnector, that provided the means of connecting to the powerdistribution cables.

Most of the transducers were delivered to Aerodyne with a K-typethermocouple imbedded in the potted assembly in contact with thepiezoelectric ceramic. This was used to monitor the level of powerthat could be delivered to each transducer. The transducermanufacturer specifies that the transducers should be capable ofwithstanding ceramic temperatures to 100 0 C. The low temperatures inthe IRT allowed much higher power levels to be delivered to thetransducers than were possible in tests at room temperature.

The tweeters had a broad-band frequency response from 1 to 40 kHz.They were also fitted with thermocouples, placing the junctions asclose to the voice coil as possible. While the temperature of thevoice coil itself was not measured, and heating is the ultimatefailure mode of these devices, the measured temperature increasedvery quickly as the damage threshold was approached. Thus operationwithin several dB of the damage power level was possible, allowing anear maximal excitation power to be used.

A microcomputer-based system provided the frequency selection andsignal generation. Software was developed for these experimentsthat selected a set of frequencies for all the transducers thatwould match their individual resonance characteristics. Thesefrequencies were then generated simultaneously on two plug-in D/Afrequency-generator boards (Quatech, Inc.); one generated the set offrequencies for the higher single-frequency transducers and theother generated a set of frequencies to be delivered to each of thebroad-band tweeters.

2.5 Transducer Power Supplies

The driving signals for all the single-frequency transducers wereamplified by a ENI high-frequency power amplifier. This device'spower output (>100 W 30 kHz-4 MHz) exceeded the sum of the powerrequirements of the transducers, but only if their impedances werematched to the 50 Ohm amplifier output. Thus the multi-frequencyoutput of the ENI amplifier was fed to a high-frequency transformerwith output for the different transducer types taken fromappropriate transformer taps (different for each type). The

19

A: R B: X A .KR. 77 500.008 Hz R: B: X o MKR 647 508.008 H:-A MAX 16.08 KQ REAL 2.3376S KQ A MAX 300.0 S REAL 28.5438 0 '

B MAX 8.800 Kq !.MAG -44.6S11 9 B MAX 400.0 Q IMAG -53.10088

A/DIV 2.088 KQ START 60 888.088 Hz A MIN 0.808 Q START 258 888.888 ?t1B/DIV 2.008 KQ STOP 100 880.088 Hz B/DIV 188.0 Q STOP 1 000 080.800 -

START-_.2S8008.08 HZ

a) (b)

A: R B: X o MKR 340 888.008 HzA MAX 788.8 Q REAL 99 9332 2B MAX 400.8 Q IMAG -16S.076 sz

ri:V-v

A M'IN 0.008 2 START 180 808.088 HzB/DIV 188.8 2 STOP 5O8 080.088 HzAMIN-_0.08980

(c)

Figure 8. FREQUENCY RESPONSE CURVES FOR THE PIEZOELECTRICTRANSDUCERS

The complex impedance (imaginary [capacitive] component isupper curve, real [resistive] component is lower curve) for thethree transducers is plotted versus frequency about the nominaldesign frequency. The individual plots represent a) 80 kHz, b)350 kHz, and C) 650 kHz.

20

amplified signal was delivered to the inductor/resistor/transducerassembly via coaxial cables and connectors.

The tweeter band (20-40 kHz) extends just beyond the audible range,so a commercially available audio amplifier was used to amplifythose signals. A Denon POA-2800 stereo amplifier was selected sinceit has good response beyond the high-frequency cut-off of thetweeter transducers. Even though both the tweeters and amplifierare considered to be expensive audio equipment due, in part, totheir high frequency response, their costs were significantly lowerthan other low frequency options.

2.6 Experimental Approach

Ice accretion experiments were performed by exposing the airfoil atthe end of the flow duct to supercooled droplet flow in the windtunnel for ten minute periods. Ice accretion thickness would thenbe measured, the airfoil cleaned, and another run started. Theoutput of the acoustic transducers was kept constant during eachrun. Transducer parameters were systematically varied from run torun. More details in the procedures followed for each experimentalrun are described below.

Starting with a clean airfoil, the IRT staff would increase tunnelflow to the desired speed and stabilize the temperature throughoutthe tunnel. This stabilization would take 5-10 minutes, withtemperature profiles measured between the water spray bars and thetest section of the tunnel being constant within I-2 0 F. The waterspray would then be turned on for precisely ten minutes. Duringthat time the airfoil was visually monitored through the viewingwindow in the control room as well as videotaped through a ceilingwindow.

Operation of the transducers was monitored electronically on anoscilloscope and via a panel meter on the ENI amplifier. Drivingvoltage was increased (via computer control) until waveformdistortion was observed as the amplifier was driven to its limit.Actual transducer driving power was checked via the temperature riseof thermocouples mounted in the ultrasonic piezoelectric transducersand on the surface of the tweeters. Temperatures of the former wereusually in the range 30-600 F. This was much lower than the ratedmaximum of 212°F, indicating that, in the cooling flow of the IRT,power levels were limited by the amplifier and imperfect impedancematching.

After the ten minute exposure time, wind tunnel speed was idled downto 10 mph and the access door opened to allow ARI and NASA personnelto enter the tunnel. Polaroid and 35mm color still pictures weretaken both of the airfoil and the transducer assembly. Figure 9shows two such pictures for ice accretion runs at 0°F and -36 0 F. Iceaccretion on the front edge of the airfoil (pointing toward the leftinto the flow duct) is clearly visible.

21

Those pictures show two extremes in ice accretion profiles. Bycomparing the two, one can see that the upper picture (a)corresponds to a thick ice layer (up to 0.6" deep). At the bottomof (a), one can clearly see the reduced ice accretion near theflow duct wall due to expected boundary layer effects. Accretionin the center section of the airfoil is somewhat uniform (seeFigure 13). In the lower picture (b) there is obviously much lessice accretion (cf. Figure 14), due to homogeneous freezing ofdroplets in the wind tunnel (see Sections 3 and 4).

Ice accretion thicknesses were determined manually by using a pairof calipers to measure the distance between front of the ice tothe back of the airfoil. Ice thickness is then given by thedifference between that distance and the 3.375" chord of theairfoil. For most experiments, the ice thickness was measured at13 points along the airfoil, as marked by the numbered stripsvisible in Figures 5 and 9.

After measuring ice thickness, the airfoil was cleared of ice.The surfaces of the transducers at the front of the flow duct werealso carefully brushed to keep them free of ice buildup. During aseries of runs, ice continued to build on the interior surfaces ofthe flow duct. This can be seen in Fig. 10, which shows aphotograph from a front view of the flow duct after an iceaccretion run. This build-up had little apparent effect on iceaccretion profiles on the airfoil. This was consistent withboundary layer calculations that predicted a stable centerlineflow in the duct.

After clearing the ice from the test fixture and closing theaccess door, tunnel speed was increased and a new icing runstarted on the clean airfoil. During tunnel stabilization, theresults of the ice accretion measurements were plotted on a laptopmicrocomputer to provide real time presentation of the data. Thetotal time required for an ice accretion run was on the order of30-45 minutes.

22

(a)

(b)

:!: ::!% ;.. .. .. .... .........

Figure 9. ICE ACCRETION ON 3.375 INCH AIRFOILThe ice accretion for Runs (a) 2-4 (00F) and (b) 4-1 (-360F) wasphotographed at the end of the 10 min run time. The minimalaccretion in (b) is attributed to freezing of the droplet cloudat the low tunnel temperature. In (a), the depletion of theice at the ends of the airfoil is due to liquid water content(LWC) losses through the duct boundary layer and deposition onthe walls.

23

Figure 10. !CE BUILD-UP IN INTERIOR OF FLOW DUCTPhotograph from front of flowduct taken after Run 3-5. Iceaccumulation on interior surfaces was clear, though continuingbuild-up had little apparent effect on accretion on airfoil.Transducer assembly was relatively clear, since it was brushedclean between runs. Ice accretion on the faces of the active(340 KHz) transducers was minimal (see p. 32,38).

24

3. ICE ACCRETION RESULTS

3.1 Research Chronology

On 17 October 1991 a kickoff meeting was held at the NASA Lewis IRTfacility. FAA, NASA and ARI personnel participated. That meetingfocused on the design of the flow duct / transducer assembly. Basedon decisions at that meeting, ultrasonic and audio transducers werespecified and ordered. Over the next two months, on-going telephonediscussions between ARI and NASA personnel led to the finalapparatus configuration schematically shown in Figure 2 anddiscussed in Section 2. NASA personnel were responsible forconstruction of the flow duct, while ARI designed and built thetransducer assembly and its associated electronics, includinghardware and computer software.

Planning also resulted in formulation of the test plan flowchartshown in Figure 11. This involves variation of IRT parameters (flowspeed, temperature, liquid water content (LWC), droplet size) andtransducer parameters (frequency, power level), depending on iceaccretion observations. As mentioned at the end of Section 1,initial conditions were chosen to optimize ice nucleation and liquidfreezing rates: the coldest temperatures, slowest tunnel speed andsmallest droplets.

The ultrasonic transducers arrived at ARI on 9 December 1991. Thatweek was spent integrating the transducers with the computercontrolled audible and ultrasonic amplifier electronics. Systemtesting focused on impedance matching of the ultrasonic transducers.On 13 December 1991 the transducers and associated electronics wereshipped air freight to NASA LeRC.

ARI and FAA personnel arrived at NASA LeRC on 16 December 1991.That day was largely spent integrating the transducers with themounting block supplied by NASA. Simultaneously, NASA personnelmounted the flow duct in the IRT and ARI personnel installed thecomputer and acoustic amplifier drivers in the IRT control room.That night a preliminary run in the IRT served to certify theaerodynamic stability of the flow duct and demonstrate rates of iceaccretion on the air foil at the end of the flow duct.

On the nights of 17-18 December 1991, a total of 23 ice accretionmeasurement runs were made in the IRT, 13 with various combinationsof transducers ON and 10 with all transducers OFF. The latter'null' runs were necessary in order to establish a baseline ofunperturbed ice accretion rates. Review of the data resulting fromthe tests of 17-18 December exhibited negative results. Discussionswith NASA personnel on the morning of 19 December resulted in thedecision to discontinue testing.

Overall, the test apparatus worked well and performed very close toplan. There were on-going problems with impedance matching

25

ULTRASONIC ICING PREVENTION EXPERIMENTDECISION TREE

Transducer power HI., all working50mph; coldest T,

vary LW.C, Droplet Size (D.S.)positive

no.effc-t* effectsuse optimal

r LWC, D.S.

Transducer power MAX, positive try each transducerrepeat, but also higher of each I n sdua erLW.C. efcsIdvdal

no effects*

go to 100mph and positiveccldest T at that speed,repe._ .atefet

no effects* -•-J switch D.S.

move transducer positiveto closest, repeatat 50mph, coldest T effects

noefecs*tweeter Select biggestfrequencies effect

use o~ptim pezotransducer closest, positive use optimal pquency100mph, coldest T effects frequency

Svary power level JUse singletweeter

frequency I vary transducer II location

*Measure accretion morphology thickness/shape var T

for small variations In a accretion phenomenaMP914-7AR.M-L

Figure II. EXPERIMENTAL TEST PLAN FLOW CHART

26

of the ultrasonic transducers which limited acoustic output powerlevels. It also became apparent that cooling from the IRT air flowcould allow driving of the transducers at power levels beyond thecapabilities of the single ENI ultrasonic amplifier. It was clearthat improvements in design could lead to enhancements in acousticoutput power levels.

In the remainder of Section 3, the actual measurements made andtheir analysis/interpretation are described in detail. Section 4discusses in more detail the conclusions and implications of thefailure to observe any acoustically induced reduction in iceaccretion.

3.2 Ice Accretion Measurements

Table 2 summarizes the conditions of the 23 runs performed in theIRT. Figures 12-14 plot ice accretion thickness (measured asdescribed in Section 2.6) as a function of position along theairfoil with zero corresponding to the transducer assembly side ofthe flow duct. Figure 15 plots the average ice accretion thicknessin the center 16 inches (4 to 20 in Figures 12-14). The x-axis hasno physical significance. The points are plotted simply in theorder in which the runs were perfc¢med. In all figures, transducerON/OFF runs are identified as labei-*d.

As labeled in Figure 14, the defining parameter for the four setsof runs was the wind tunnel temperature, which was varied between00 F and -36 0 F. In accordance with the left ('no effects') side ofthe test matrix (Figure 11), liquid water content (LWC) and flowspeed were also varied, as listed in Table 2 and labeled inFigures 12-14. The chronology of the different runs is describedbelow.

Set 1 was acquired on the night of 17 December at a tunneltemprature of -20 0 F. The first four runs were at 50 mph, the firstand last (1-1 and 1-4) with the transducers OFF. The next five runswere at 100 mph flow speed, with the first (1-5) again being runwith the transducers OFF. Since there was no significant change inthe ice accretion , it was decided to run at the slower 50 mph forall subsequent runs to maximize time available for droplet freezing(see Table 1).

Likewise, the results of Set 1 established the pattern oftransducer power levels for subsequent runs. Run 1-2 distributedacoustic power among 20, 25, 30, 35, 40, 340 and 650 kHzfrequencies. At that time it became apparent that impedancemismatches interfered with simultaneous operation of the ultrasonictransducers. In view of the lack of any acoustically inducedeffect, it was also decided to concentrate the power by driving thetweeters at one frequency. Thus, in all subsequent transducer ON

27

0.8 1 1 1 1

........ ON TransducersIOFF

0.6- 5 MPH._2"LWC: 0.9 g/m3

£ '""" ...............

CL0 ............................0S0.40

<• 100 MPH

•0- 2- LWC: 0.6 g/m30.2

SET 1: -20°F- MVD: 14 gmI

0.0 I I I I0 5 10 15 20

Distance Along Airfoil (in)

Figure 12. ICE ACCRETION PROFILES FOR DATA SET 1

The accreted ice thickness is plotted at 1 inch incrementsalong the airfoil span, measured from the duct wall in which

the transducers were mounted. Depletion at the edges is due to

boundary layer effects. The first three profiles only extend

to the midpoint, after which the possibility of profile

asymmetry prompted taking meaFuirements along the full airfoil

span. The 100 mph data, at reduced LWC, corresponds well with

the 50 mph data and the lower speed was used for subsequent

runs to maximize the time available for droplet freezing.

28

0.8

ON TransducersS........ OFFI

0.6 .

.. .....................

S0.4 -

SET 2: o°F50 MPH

LWC: 1.35 g/m0.2 MVD: 40 gm

0.0 I I I0 5 10 15 20

Distance Along Airfoil (in)

Figure 13. ICE ACCRETION PROFILES FOR DATA SET 2The lowest profile, with the transducers OFF , was the firstrun for these conditions and apparently represents non-equilibrium conditions in the tunnel. Variations in thesubsequent runs indicate the reproducibility of thesemeasurements.

29

Q)u

0 a O 9CN r-4 w 00U)ko 0 UL'1 OD l m e' Ul to -cve r-4 ko 1.0~'0,') U) -4 r- U~. NrN 00C D r or - r- nU o a D-4LA' N U) r- 0) LA

to ...4 C

r, 4 ~ r24 4L. r2 k DZ 4 UZ. rx.0 0 0 0 0 0 0 0 0

11 1 -4 1 1 fn W I I I I I W I0

II, - . I ý4N 4 1 I 1 1 I I I . I I ko0 I I I ý

0 . 1 am ) I I I I 1 1 -I I I-

-i U4 Ln-U) 9 ý

-Z 0

LOA

3 1 4 I 1 _ I I I I I I I f-4() I I I I I 4 I I o4

U) LAU)U

C I N 4 I I .4 I I OI I 1 m i C- I I I, 1 I 1 1 I-

E-4

0 0-0

0,~

-0In In In a

U .ALAU~to0 en N~ (n4.0 p LA 41' U D~~~~~.4~~~~~~~: U) ~t4 AVr40~.4y4 A .0'4 mL~4a

J.) *.. '.0 LQ~~~ LA.,IDN N O ~ .44 '

N a,

4- --4ý

1 30

0.8

I • ISET 3: -3*F

0.6

CL

a) 3oLWC: 1.35 glm

C 0.4 MVD: 40gm0

ON Transducers..... ..OFF

0.2C * SC. :.. :.iiii ....

[.ET :-.F.............

0. 0L.0 5 10 15 20

Distance Along Airfoil (in)

Figure 14. ICE ACCRETION PROFILES FOR EXPERIMENTAL DATA SETS 3 AND 4These experiments were run at the coldest temperatureachievable in the IRT . The controlling software wasprogrammed to take the tunnel to -30*F but, in fact, overshotto -32 0 F for the first Lun of set 3. After warming to a true -

300 F, the software was modified by IRT operators and -36 0F wasachieved for run 4. Such cold temperatures had not beenpreviously maintained in the tunnel with its current systemcontrol unit.

31

runs the tweeters were driven at a single frequency (10, 20 or 40kHz) while driving one pair of the single frequency ultrasonictransducers.pDuring the last run (1-9) on 17 December the tweeterpower was increased to 80 watts at 10 kHz. Sometime during the runthe tweeters failed. This was consistent with maximum power levelsspecified by the manufacturer. Thus, the tweeters were replacedduring the morning of 18 December. Also, it had become clear thatone of 650 kHz transducers was emitting low power, probably due tosevere impedance/resonance mismatch with the other. Thus, thattransducer was replaced with the 650 kHz spare, which did indeedperform better.

The performance of the ultrasonic transducers could also bemonitored by the amount of ice accretion on their front surfaces atthe end of a run. The transducers that were ON were virtually clearof ice. In the case of the defective 650 kHz transducer, theimproved performance of the replacement on the second night waseasily observed from reduced ice accretion on its front surface. Atemperature rise of -10 to 15 0 F was measured on the transducermounting plate surface after one run. It is not clear whether this'anti-icing' was simply a heating effect or due to the ultrasonicvibration of the surface or some combination of both. However, itserved as a useful de-icer in these tests.

On the night of 18 December, LWC was increased to 1.35 g/m3 , whilemaintaining the slow 50 mph flow speed. For Set 2, tunneltemperature was set at 00F. The first run (2-1) was with theacoustic power OFF, providing unperturbed ice accretionthicknesses. Runs 2-2 to 2-4 sequentially powered the differentpairs of the ultrasonic transducers and different drivingfrequencies in the tweeters. Run 2-5 again was a transducer OFFrun. The unexpectedly low ice accretion for Run 2-1 in Figure 13was the first run of the night and was attributed to unequilibratedtunnel operating conditions.

The variation between Runs 2-1 and 2-5 emphasized the importance ofrunning as many OFF as ON experiments in order to measurevariability of ice accretion under 'constant' icing conditions.This was particularly important in view of the negative results,i.e. it was necessary to establish how small an anti-icing effectcould be detected.

For Set 3 the tunnel was cooled to -30 0 F. Run 3-1 was again a nullrun. Starting with Run 3-2 and then with 3-4 and 3-5, theultrasonic transducer and tweeter frequencies were againsequentially powered. Since ice accretion during Run 3-2 was somuch deeper than that in 3-1, another OFF run (3-3) was immediatelytaken which showed that Run 3-1 was anomalous. At that time, itbecame clear that, due to a temperature programming error in the IRTcontrol software, that Run 3-1 has actually been about 2'F colderthan the -30°F target temperature. The shallow ice accretionobserved at -32 0 F was the first indication of spontaneous freezing

32

of the droplets in the wind tunnel as tunnel temperatures neared thehomogeneous freezing point of pure water(-400F)

This was readily apparent when the tunnel was cooled to -36'F in Run4 as displayed in Figure 14. Ice accretion thickness decreasedsharply (to - 0. 1 11 average thickness) , presumably ref lecting thedegree of spontaneous freezing of the droplets in the tunnel,independent of any added perturbations. Due to lack of tunnel timeat the end of the evening, only three runs could be recorded, twoOFF (4-1 and 4-3) with one ON (4-2). There was significantvariability in the observed ice accretion thicknesses; the degreeof spontaneous freezing is very sensitive to the precise operatingconditions of the tunnel. Looking at the relative thicknesses inFigure 14, there is no indication of a (statistically) significantreduction in ice accretion for the transducer ON Run 4-2.

3.3 Summary

The ice accretion/acoustic transducer testing apparatus performedvery much as planned. The IRT facility operated nearly flawlessly.During the first day (16 December) all problems concerning assemblyand integration of the transducers and flow duct were resolved.Furthermore, integration of the computer control system and thetunnel control room facilities went very smoothly. NASA provided anIBM compatible PC to speed up software control of transducer drivefrequencies. During actual testing, NASA personnel were able toprovide exactly the super-cooled cloud conditions requested.

The flow duct/airfoil design provided well-defined ice accretionconditions. Ice thicknesses on the order of 0.5m in 10 minutes oftunnel time were ideal for anti-icing testing. The ice profileswere uniform (for the most part) and reproducible. The profiles doshow some asymmetry which appears to be a property of the ductitself, uncorrelated with transducer operation. The fall-off oficing at the edges is consistent with expected boundary layereffects. Overall, it appears that the flow duct served its purposeof providing a reproducible well-defined center line flow, asdemonstrated by similar accretions on the downstream airfoil.

The transducer assembly also performed as designed, though it wasapparent that, in the case of the ultrasonic transducers, higherpower levels could be achieved with better impedance matching andmore electronic amplification. The problem was the lower operatingtemperature and increased cooling in IRT. The lower temperaturesshifted the resonant frequencies of the transducers, invalidatingmuch of the impedance matching/ resonance calibration testingperformed at ARI the week before at room temperature. Furthermore,in the wind tunnel flow it became clear that the transducers couldbe powered beyond the capacity of the ENI amplifier.

33

As summarized in Figure 15, the overall result of the testing isthat the ice measurements show no evidence of any reduction ofaccretion by the acoustic power. Over a wide range of tunnelconditions (e.g. temperature, speed and LWC), there is nostatistically significant difference in ice accretion due to thepresence of ultrasonic energy. The desired reduction in iceaccretion simply was not observed. Beyond the ice thicknessmeasurements plotted in Figures 12-15, direct examination of theaccreted ice showed no indication of any changes in ice morphologywith the acoustic power ON.

The variation in ice accretion thickness is seen most clearly inFigure 16, which plots ice thickness versus temperature. Iceaccretion thickness has been normalized to the liquid water content(LWC) in the IRT. The normalization places data from Set 1 on thesame relative scale as Sets 2-4 This normalization is notrigorous, since one should also include the collection efficiencywhose estimation would requires detailed air flow calculations.This would have the largest effect on Set 1 (at -20 0 F) whose smallerdroplet size (14 versus 40 gm MVD; see Table 2) would reduce thecollection efficiency. Thus, the plot in Figure 16 probablyunderestimates the relative ice accretion thickne.: at -20 0 F.

Figure 16 indicates that ice accretion on the streamline airfoilpeaks at about -20°F, perhaps reflecting a maximum in stickingprobability for droplets supercooled at that temperature. Moreimportantly, below -20OF ice accretion drops off, especially below -30 0 F where it drops sharply. This is exactly what one expects asthe homogeneous freezing point is approached (for example, seereference 1). This sharp drop off clearly explains the shallow iceaccretion observed in Run 3-1, when the IRT temperature was notbeing properly regulated. The plot in Figure 16 implies that thetunnel temperature may actually have been lower than the -31.9 0 Fthat was extracted from the ESCORT thermocouple readout printed atthe end of the run. Unlike every other run, regulation/monitoringof temperature by IRT operators was unreliable during Run 3-1 due toa programming error which was detected between Runs 3-1 and 3-2.

Most importantly the conditions for Set 3 and 4 were optimal forobservation of an acoustically-induced ice nucleation effect. Thelow ice accretion is evidence of the desired anti-icing effect:droplets that freeze before impinging on the airfoil bounce ratherthan stick. The temperature scale of Figure 16 is very consistentwith fundamental cloud microphysics. The sharp ice fall-off withtemperature indicates that the remaining unfrozen supercooleddroplets are just below threshold for ice nucleation. That is themost sensitive point at which to observe a perturbation of icenucleation probability. Thus, the lack of any indication of anacoustically-induced ice accretion reduction represents a clearnegative result.

34

0.8SEET 3

0.6- SET 1-• 20O 0;• F

o %

o 0.4-

0.2-

0 ON :Transducers

0.0

Figure 15.AVERAGE ICE ACCRETION THICKNESSES FOR THE CENTRAL AIRFOILSPAN

The ice accretion in the central region of the airfoil awayfrom the depleted edges is averaged for all runs in each of thedata sets and is plotted in the order the data were taken.

35

0.8 1

0 ON Transducers

S0.6

0

S0.4-

Zw

00a:

o 0.2 0O4

0.01 1 1

-40 -30 -20 -10 0 10 20 30

Temperature (OF)

Figure 16.NORMALIZED ICE ACCRETION THICKNESSESThe data in Fig. 15 are divided by the liquid water contentcorresponding to the individual run and plotted versus tunneltemperature. It is apparent that spontaneous freezing isreducing the normalized ice accretion thickness in the range of-25 to -36*F and that the transducers do not have any effectabove the experimental uncertainty. As discussed in the text,the normalized thickness for the Set 1 data at -20 F probablyunderestimate the relative ice thickness.

36

Comparison of the conditions explored in Table 2 with the testplan shown in Figure 11 shows that the tests followed the plan onthe negative effects (left) side of the diagram. The one flaw inthe overall test procedure was the use of high (1.35 g/m 3 ) LWC inthe tests at the coldest temperatures on the night of 18December. Unfortunately ARI personnel did not realize at thetime that this high LWC had nearly tripled the average dropletsize (40 Am versus 14 Am MVD at lower LWC). This is significantin view of the freezing times listed in Table 1. Realizing thattF scales as the diameter squared, for 40 Am droplets at -30°Fthe predicted freezing distance at 50 mph is 1.2 meter. Whilethis is shorter than the 2 meter distance actually between theirradiated volume and the airfoil in the flow duct, expectedresults for the initial tests were to observe the smallestdroplets possible to maximize freezing rates. For that reasonand in view of the negative results, the transducers were notmoved forward to check shorter freezing times.

As mentioned above, no further tests were undertaken after 18December. Optimal supercooled conditions had been tested and thenegative results did not indicate that further testing could bejustified as a productive use of NASA IRT resources.

37

4. DISCUSSION

The goal of this research project was to demonstrate thatacoustically induced nucleation of supercooled water droplets canresult in freezing the droplets. To this end, a total of nineacoustic transducers were specified, mounted and coupled tospecially designed flow duct and airfoil, which were placed in thecenter of the NASA LeRC IRT facility. The design, assembly andintegration of this apparatus, done collaboratively with NASA andARI personnel, was accomplished almost exactly according to plan.The operation of the transducers and the IRT during testing wentwell. Unfortunately, no acoustically induced effects on iceaccretion were observed.

In the remainder of this section, the results in terms of theirimplications for ice nucleation (and icing prevention) and ofpossible reasons for the negative effects are discussed. Thisincludes possibilities for improving and extending the experimentalprogram. ARI concludes with a summary of conclusions andrecommendations.

4.1 Ice Accretion Observations

The ice accretion profiles (Figs. 12-14) measured for various cloudconditions and temperatures were seen to depend on the tunnelparameters. The maximum ice thickness along the airfoil span variedfrom about 0.1 inch to about 0.6 inch as the temperature, dropletsize, and liquid water content were changed. The ice accretionthickness also varied along the span, dropping to smaller valuesnear both ends and sometimes having a definite asymmetry end-to-end.The end-to-end asymmetries do not correlate with transduceroperations nor with mean tunnel conditions: temperatures and cloudconditions. The most likely source for such variations are localnon-uniformities in the liquid water content of the spray cloud,whether due to the tunnel as a whole or due to perturbations causedby introducing the flow duct in the test section. The greatestmeasured asymmetries are larger than expected from the reported LWCuniformity profiles in the IRT Users Manual (Ref. 10) although, forother runs, very good uniformity was observed.

The reduced and integrated profiles (Fig. 15), representing thetotal accreted ice away from boundary layer effects, demonstratesthat all of the variation in ice thickness is attributable to tunnelconditions. The decrease in ice thickness at the ends of the spanare due to droplets depositing on the internal flow duct walls asthe sampled flow convects toward the airfoil. Deposits on theinternal walls were observed and required cleaning where they mightimpact the transducer operation. It is interesting to note that,when any of the transducers or tweeters were operating at high powerlevels, their emitting surface did not accrete ice (except for the350 kHz, which always accreted at least some ice). Given the modest

38

temperature rises -n their emitting surfaces, this effect seems tobe due to solid a- stic wave motion interfering with the iceaccretion process.

The only variations remaining after normalizing by the cloud liquidwater content (see Figure 16) is a systematic variation withtemperature and run-to-run variation that cannot L-e correlatedsystematically with transducer parameters. The temperaturedependence shows an increasing normalized ice accretion as thetemperature decreases until spontaneous freezing begins below -20'F.Ice accretion is nearly totally suppressed due to spontaneousfreezing at -36 0 F. The run-to-run variation does not correlate withtransducer frequencies being used and appears to be wholly due toexperimental uncertainties, either in tunnel conditions ormeasurement errors. These results indicate that, over the range offrequencies and power levels used to excite the super-cooleddroplets, acoustic suppression of ice accretion does not occur. Theice accretion measurement used is sensitive to complete and partialfreezing of the droplet cloud as is evidenced by the drop in iceaccretion as spontaneous freezing progresses. Various degrees ofcloud freezing resulted in the decreasing ice thickness asspontaneous freezing progressed into the -25'F to -35 0 F range.

If acoustically induced nucleation had occurred, partial freezingwould have happened almost instantaneously and continued until thedroplet temperature rose to the freezing point. The only way thatnucleation could have happened and not resulted in a measurabledifference in ice thickness would be if ice accretion was notaffected at all until the droplet was completely frozen. In orderfor this situation to have occurred in the present experiments, theevaporative cooling rates for the droplets calculated (See Table 1in Section 1) would have to be in error by almost an order ofmagnitude. For these reasons, it is believed that no nucleation wasinduced with the ultrasonic levels employed in these experiments.

4.2 Acoustic Power Levels

The total acoustic power within the excitation volume dependson he power delivered to the transducer and its coupling to theacoustic radiation field that leaves the transducer surface. Thesingle-frequency transducer manufacturer and the tweeter supplierboth provided specifications for optimizing the coupling efficiency.The single-frequency transducers are impedance matched with air in aquarter-wave coating on their active surface. Acoustic coupling wasmeasured for all the transducers used in this study and was found tobe consistent with the manufacturers, specifications. While thesecouplings are considered quite high for the coupling of solid-acoustic waves into air and for audio equipment, only a smallfraction of the total power supplied to the transducers actuallypropagates into the duct.

39

The acoustic coupling losses for each of the transducers arelisted in Table 3, along with the sound pressure level (SPL) in dBfor a representative input power level. The lower tweeterfrequencies combined with the smaller tweeter acoustic apertureresult in significantly more diffraction of the energy for thesecases. The table entry for the tweeters is for the SPL at thecenter of the duct, with higher levels near the aperture and lowerlevels further away. The higher piezoelectric frequencies do notdiffract significantly for their transducer sizes and wavelengths.

Reflections, which may make local levels higher and may alsointroduce standing waves and nodal patterns, are not included inthese estimates.

Table 3. Acoustic Power Levels

Frequency Coupling Input SPL(dB)(kHz) Losses(dB) Power(W) at 30 cm

20-40 32-35 15 11580 35 50 125340 35 50 125650 35 50 125

By increasing the coupling efficiency further, more acoustic powerwould be available to excite the super-cooled droplets in the flowmore strongly . Increases in coupling efficiency could raise theacoustic power levels by from two to over three orders of magnitude(recovering 20 to 35 dB coupling losses of the various transducers).For any realistic impedance matching, residual losses will preventcomplete recovery of all the coupling losses, but substantialimprovements beyond the performance of the current transducer set isconsidered possible by the manufacturer (NUTRAN).

4.3 Oneratina Parameters

The existing empirical evidence for acoustically induced freezinghas all been obtained in natural clouds or in cloud chambers wherefreezing times are not limited. The effective freezing times in thepresent experiments, for 50 mph (22 m/s) tunnel speed and less than2.4 m between the transducers and the airfoil, all are below about0.1 sec. It is conceivable that crystallization nucleation hadbegun even in the experiments reported here, but in this flowsystem, there had not been sufficient time for the crystallizationto propagate throughout the individual droplets. If the heat of

40

fusion cannot be removed from the freezing droplets quickly enough,as described previously, completion of droplet freezing will becomekinetically limited.

An alternative means of exploring the detailed mechanism responsiblefor the previously observed freezing would make use of a cloudchamber to repeat the conditions of these earlier experiments.Power thresholds, frequency dependencies, and the temporal behaviorof the freezing process could then be measured quantitatively andexperimental conditions for a successful icing wind tunnel studycould be calculated. Cloud chamber facilities at Cold RegionsResearch and Engineering Labs in Hanover, New Hampshire and at theColorado State University at Fort Collins (See Reference 1) havebeen canvassed and both have facilities where such studies could beperformed.

While no previous experiments have quantified the power levels orfrequency dependence of those empirical observations, the vanStratten-Allen patent 6 refers to super-audible frequencies (theyquote the range of 15-30 kHz) at power levels above 150 dB. Theirpatent grew out of cloud chamber work where freezing times were notlimited by flow times. They do not provide evidence that 150 dB isa SPL threshold for acoustically inducing droplet freezing. Verbalcommunication with F. van Straten indicates that the threshold SPLlevel was not measured in any detailed way, but it is worth notingthat the SPL levels used in the present experiments are 25-35 dBlower than this value.

While the means of increasing SPL levels in the high frequency range(above 80 kHZ) is probably best accomplished with improvements inthe impedance matching of the piezoelectric transducers, analternative transducer may be required to increase the SPL in thetweeters' frequency range. The van Straten-Van Allen patentproposes using a siren operating in the 15-30 kHz range at a SPL of150 dB and electronic sirens are available commercially that roughlycorrespond to their suggested acoustic driver.

At this point, it should be noted that simple utilization of such acommercial siren (or increased transducer amplifier capability)probably would not be practical in actual icing prevention systems.A 30 kHz operating frequency may be too close to the audible andpower requirements may be excessive. The key point is that, givenpositive ice freezing results, observed under any conditions, therewill be ample room for engineering improvements to design andconstruct a realistic device.

4.4 Summary

The basic conclusion to be drawn from these anti-icing tests isthat, for the range of ultrasonic power levels (-125 dB) andfrequencies (20kHz to 650 kHz) used with the flow times (-100 ms)provided for droplet freezing, no measurable effects on ice

41

accretion were observed. The tunnel temperature was varied,including temperatures low enough (-36 0 F) for spontaneous freezingto occur independent of the ultrasonic excitation, and thetemperature dependence of ice accretion was observed. Particularlyat the lowest temperatures, ultrasonic induction of freezing wasexpected to affect ice accretion if the operational parameters wereabove the threshold for nucleation.

Given that anecdotal evidence does exist for acoustic (both audibleand ultrasonic) induction of freezing, it appears that the currentexperiments do not have a large enough operating envelope to capturethe relevant processes. It is important to note that the presentresults report quantitative experimental conditions for acousticexcitation of cloud droplets for the first time. They provide a setof baseline conditions for any further exploration of the parameterspace of acoustic power, excitation frequency, and freezing time forthese type of cloud conditions. The induction of freezing appearsto be a threshold phenomenon, and is undoubtedably highly nonlinearin any case.

Expansion of the parameter space (or operating envelope) in anyfuture experiments should be focussed on 1) increasing transducerpower and 2) increasing the freezing time. While frequencies couldalso be varied further into the audible range and further into theultrasonic range, the broad range used in the present studiesextends from where attenuation in air becomes important at the shortwavelengths up to several millimeters, much larger than the dropletsizes. The transmitted transducer powers were below the referencedpower level of 150 dB in the van Straten-van Allen patent. Theselevels could be increased by engineering improvements in thetransducer designs and/or an alternate acoustic source could beconsidered. Commercially available electronic sirens havespecifications to higher power levels than used in the present studyand their use in future experiments would be recommended ifappropriate frequency response could be obtained.

The addition of a second ENI high frequency amplifier would doublethe available ultrasonic transducer power. It would also enableseparate frequency tuning and impedance matching of each ultrasonictransducer pair. This, together with the siren mentioned above,would do much to close the gap between the current 125 dB powerlevel and the 150 dB level specified by Van Straten.

Greatly increased freezing times are not possible in the NASA LeRCIRT. The tunnel was running at the lower end of its speed range andthe flow duct could not be extended significantly without having theboundary layers affect most of the flow cross section. The mostfeasible means of increasing the freezing time would be to performsimilar experiments statically in a stationary cloud. Severalpossible laboratory facilities exist for such experiments.

42

5. Conclusions

1. No changes in ice accretion rates (i.e., no icing prevention)due to acoustic perturbation were observed.

2. Acoustic power levels (up to 125 Db) were about as high asplanned. However, future enhancements (better impedancematching, increased power amplification) are possible based onIRT experiments.

3. Failure to observe icing reduction may reflect non-linearity ofice nucleation process; experimental parameters may still liebelow operational envelop (nucleation threshold).

4. Convection times in the icing tunnel may not have been longenough to provide sufficient droplet freezing.

5. This series of tests provides the first systematic baseline forevaluating anecdotal evidence of acoustically inducednucleation.

6. The anecdotal observations of acoustically-induced nucleationshould be understood systematically by expanding thefrequency/power/freezing time operational envelope.

7. It is important to note that, once that the operational

envelope is characterized, ultimate feasibility of aircrafticing prevention may depend on engineering development intransducer design.

43

6. REFERENCES

1. DeMott, P.J. and Rogers, D.C. (1990) "Freezing Nucleation Ratesof Dilute Solution Droplets Measured Between -30' and -40'C inlaboratory Simulations of Natural Cloud," J. of Atmos. Sci. 47,1056-1064.

2. B.J. Mason, "Clouds, Rain and Rain Making," p. 69, CambridgeUniversity Press (1975).

3. S.S. Hirano, "Ecology and Physiology of Pseudomonas Syringae,"Bio. Tech, 1, 1073 (1985).

4. R.L. Ives, "Detection of Supercooled Fog Droplets," J. Aeron.Sci., p 120 (1941).

5. W.C. Swinbank, "Crystallization of Supercooled Water byUltrasonic Radiation", J. of Meteoro., 1A, 190 (1957).

6. F.W. Van Straten and W.A. Van Allen, "Method of CrystallizingSupercooled Water Droplets",U.S. Patent No. 2,480,275 (1949).

7. D.R. Worsnop and Z. Hed, "Development of Icing Prevention byNucleation of Ice Particles in Front of Subsonic Aircraft",Final Report, ARI-RR-858, DOT Contract No. DTRS-57-90-C-00132(1991).

8. M.S. Zahniser, D.R. Worsnop, C.E. Kolb, J.M. Van Doren, L.Watson and P. Davidovits, "Heterogeneous Reaction Kinetics ofImportance to Stratospheric Chemistry", Final Report, ARI-RR-674, Chemical Manufacturers Association, Contract No. 87-653(1988).

9. D.R. Worsnop, M.S. Zahniser, C.E. Kolb, J.A. Gardner, L.R.Watson, J.M. Van Doren, J.T. Jayne and P. Davidovits,"Temperature Dependence of Mass Accommodation of SO 2 and H20 2 onAqueous Surfaces," J. Phys. Chem.93, 1159-1172 (1989).

10. Soeder, R.H. and Andracchio, C.R. "NASA Lewis Icing ResearchTunnel User Manual", NASA Technical Memorandum 102319 (1990).

44

APPENDIX A

METHOD OF CRYSTALLIZINGSUPERCOOLED WATER DROPLETS

F.W. Van Straten and W.A. Van AllenU.S. Patent No. 2,480,275

1949

45

Patented Aug.'30, 1949 2,4809275

UNITED STATES PATENT OFFICE

51ET1IOD OF CRtYSTALLIZING SUPMRCOOLED WATER DROPLETS

rlorence W. ran Stralten. Green AceLa Md.. anWlllilam A. Van Allen. Beaten. Mass.; said Van..Jlea assigner to said van Strains

No Drawing. *Appllieste October 4. 19417.Serial No. 778.020

4 Claimns. (CL 244-134)

our invent~on relates to meteorology and coin- We have discovered that sound Waves of prop42pr~imc a method of cryitall~r~ng super-cooled frequency and Intesity may be focused in a beamwater droplets lying In the path or -,n airplane and used to produce almost Instantaneous Crysta-in order to pfesent the formation ef Ice thereon. Ilainu of super-cooed water droplets. 13y

The most Important object of our invention 5 mounting a suitable sonic generator and radiator.Ms to IMVMove the VtdetYI with which aircraft or horn, upon a plane we can project a soundoperations may be conducted. beam ahead of thepa e 1:11n18md crytallize the super-

Another Object or the inventuon is to increase cooled dropilts lying abead of it. The ice crys-thte milietry efficiency of ai~r-rnft, p-trucularly tois then either bounce off the plane or flow byin frigid regions. 10 it in the air stream. Consequently ice does not

A furtlicr Object of the invention Ii, to mrvlde build up on any Of the surfaces thereof.I'rst~cnl compnet. mcenns to be carried by P. plane We befleve that the effect Of sound waves on',nd useful to eliminate the hazard of ice formna- the super-cooled droplets is two-fold. Agce=-tUo' on vwtngls and other exposed surfaces. penyin the sound waves is an adiabatic tam-

Aliti~iOft rotWet ef 's sthorol.!Oglcsi Phinic'- 15 pmrt=r Change: at 150 db L R.X S. piXNOena involved are not too clearly understood. it Is leve this change is .051 C.. but at 160 db. theknew"t that t"aot -w'here may7 contain, in cer- change is LW C. Com1111111setamy tile sudden teai-ta'zs yrone". wa Cr droplets In a metastable. super- peratune drop may. In certain cumasi.st to in-cooled conditio at temreratures as low as -3&00 Crease random crystallizationl to the point wherec. It has been shown expe~rimentally that ice 20 substantially all of the droplets in the iminedi-erystals and sup~er-cooled, water droplets may sin Vicinity wial rapidly crystahlim.exi st smultanteausly in clouds where the upwsrd Furthermore the pymsure gradient accom-comrponent of air velocity is sufficient to effect panying the puusressive sound wave brings dis-towe prodL'wt'on of super-cooled droplets faste tortingstresses to bear onthe dropitts.9asin dils-

t''ntey can evoaorate over to nearly Ice crys e3 turbing their condito of equilibrium and effect-mix. ig almost Instantaneous 0r7stalllztiOn thereof

Sno~re erystallf2ation of the droplets is brought Inasmuch as the propellers of a Plane emit Aat the revt'lt of colWsons due to Brownish considerable volume of sound at vai~ous [fre

mnovement of t"e droplets: the crystallization quencles without affecting the formation of fee.t-'kg'q places on doist motes or other nticlei present 30 i$ Is obvious that sound per se is not sufficienitin the cloud. The rnte at which ituch random and that frequency and Intensity imust be esat-erytailltaltlon occin~sil clotely dependent ýin tern- fully selected If the desixed results are to be40ecraore P~vr evampl, the ritte at -35.0 C. is toned on a sastsfatoem SCaIl.ton5 r~s ,%reat as at -34.11' C. If the tempprat- TO maw funl Use Of tae pressize wave aspewtturp falls sianisfleantv below -33.00 C.. ubstan- 33 of the sound wave the frequency must be suchtie'ly all of the drn~letx wilt crysallIze. %tat, there "I be deuitsM WO m Oioof %b th v" wt

11oreover. it a stupe-r-ooled droplet is subjected respec to the droplets. Below 1 kc. the dleVM~lto dlstetW1111 streame as by in Intense amend wave will move with the si wit alhmalt no relativeor by Import on t movin; 'urface. its moetstabl Movement between individual drops. Above SOconditt'w nr eattillbulutm in dliturbed, and It turn 40t Ic. the attenuationi of soUnd energy Is pent. andP',otr at onwe to Ice. This is the nanner In the necessary intensity Must be Maintain" atwbfe'9 ire fnrm~s on aircrft wings and the ll.. sufficient distanc ahead of the Plane to give the

Keretefore the problem of offiminating ire droplets time to crysta~lliz before being St-formuation on aircraft has been d~rected sloan by the plane. cIniseuntl. 30 kc. would ap1-three avenues of attack. One system commcsi -13 peair to be a preferable uVppe limit Of frefleliisomrious reailient boots for Protection of vital gfoweve. since the dulled effects may be Obh-,urfseepm: aPer Ica fovm'i on the boots. the pilot tsigied at frequencies up to 100 kc.. the scoe Offlesm then to crack off I he Ice. Another method our invention is not Unted to 30 kc. as an UPPUpt to conduct a hot fluid to the vicinity of the limitLritrface to be proteftod In arider to molt the Ice 51 Inasmuch as soW Is MOOe esl focue as

asit fa.R Chemical compounds have boon aps- Its frequency WnrAsas "An since a beam Is des*,i'4led ato suc surfaces In order to reduce their Sime In order to concentate the avsailbl e'adhogivte enutlaUs and thereby in hlIt ice forma- cra In the most useful areas we find that atili. Kowever. no one thus far has oreducid freiquency Of frOMISt 1 .36 111% is 11111ae.1M in-amS Gen wa shlafommy soutio sote 4" ee. 85 I emesal. mesfaVSinies athi ranli 10 that

46

ItHoIWVth auibe mILCoseqenlyan Ac~ting the droplets and surrommUn air to theInst allation of equipment for Performing wei aCUMo of sound Wave having an inteunsiy ofrethod of the Invenuon does not consutitte a at least 150 db. and at a frequency of from -m tonuisance from the standpoint of noise, one hundred kllocycles. which freqomeuy and ill.While the apparatus used fornis no ;iart per 6 tenuaty will be stumelent to caume vibration w"*cNo of our Invention, we suggest the use o( a i1ower- impart 3uMclent, distortional stresses withifanthful siren capable of producinx sound at an in- dropleta to crystallize the same.

tensity of better than 150 dli. Said at a frequency 3. A method for crystallizing super~cooaegin the range 15-30 ke. The siren may be coupled Watter droplets suspended In air. comprising sub.to a raditinug horn serving to focus the 3oand 10 Jecting the droplets and surroturnin air to taefrom thif siren Into a fan-shaped beazri. By action of sound waves having an Intensity of atmounting the horn at a ccftial Point on the least 10 db. and at a frequency above the sudobiPlane We are able to projecita sonic beam along limit but lesw than 100 ke., which frequency andthe natural drection or movement of the plane intensity will be sum lcent to cause vibration whichand thereby effect cryxtallization of suswreoowe I r, mpart sufficient, distortional strenses Within thewater droplets lying In Its path. The beam Is1 droplets to crystaize the sae.tso radiated as to effect, crystallization through 4. A method of Protecting aL moving surf seetan area ling sullclently far athead or the plane against ice formation. comprising cr7saimiwngto Permit complete crystallfzation and broad supRO-cooled water droplets lying in the pasthenough to eXten completely across the width 20 of movement by projecting a sound beam aheadof the plane. of and in the direction of travel of the surfaceHaving thus disclosed our Invention, what we the sound beam comprisin sound waves at aclaim as new and desire to secure by Letters frequency between 15 and 100 he. at an intenityPatent of the United States Is: of at least 150 db.. w~ich frequency and intensity1. A method for preventing Ice formnatiuon ofl.! will be sufficient to cause vibration which imparts;moving surfaces. comprising crys3tallizing super.- s"911clent dllstort-ional stresses within the dropletscooled water droplets lylg in the Path of flight to crystatllIze the same.by Projecting a sound beam ahead of the moving PWRJ29CE W. VAN STRATEN.surface and through a cr055-sectional area at WTLLL46M A. VAN ALLEN.least as wide ats said surface,. said sound helm 30comprising sound waves at a frequency of 15 RVWFRNCES CITEDke.-30 ke. and an intensity of lirreater than 150decibels& which. frequency and intensity will be The following references are of record in thesufficient to cause vibration which Jimpars XU. Mle Of this patent:clent distortional stresses within the droplets to 3S UNITED STATES PATENMIcrystallize uthese2.L A method for crystallizing super-cocied Number Name Datewater droplets suspended In air comprising sub- UIA.171 Amy------------- Nov. 13. 1134

47

APPENDIX B

ESTIMATE OF POWER REQUIREMENTS FORULTRASONICALLY INDUCED ICE NUCLEATION

48

APPENDIX B: ESTIMATE OF POWER REQUIREMENTS FOR ULTRASONICALLYINDUCED NUCLEATION

Energetics of ice nucleation involve consideration of the surfaceenergy of embryonic nuclei, about which little is known. In onelimit, the surface energy density (E,) required to form an embryonicnucleus (of radius r) can be approximated by the heat of fusion (Ce)of its volume, which gives an energy density of

E, = 10-3 joule/cm2

This formulation estimates that the energy barrier to formation ofan ice-like surface within liquid water is equal to the energy gainin forming an ice crystal. This is not an unreasonable assumptionfor molecular rearrangement processes; it probably gives an upperlimit to the barrier. The actual magnitude of the barrier dependson the detailed mechanism of the water/ice transition.

Another approach to estimating surface energy is to consider theenergy required for bubble formation. In other words, consider theultrasonically induced ice formation as a cavitation nucleationprocess. For small droplets (for which the heat of vaporization ofthe interior volume is negligible), the surface energy density issimply given by the surface tension of water:

Es = 75 dyne/cm - 10-5 joule/cm2

This smaller value clearly indicates that water surfaces can beperturbed by energy densities far lower than the upper limit statedabove. Although not convinced that bubble formation is required forice nucleation, ARI will use this energy density estimate as a morerealistic estimate of the energetics of liquid perturbations.

The next question is the time in which this surface energy needs tobe delivered to the droplet. In other words, what is the powerdensity required to produce embryonic nuclei for icecrystallization? This can be estimated by considering nucleusformation as a molecular rearrangement process, whose rate can beestimated by the rate of molecular diffusion within the liquid.Then, the time scale is given by

t = r2/D1

where D1 is the diffusion coefficient of water. For D1 = 10-5 cm2 /sand r = 10-5 cm, one obtains

t = 10-5 s

Combining this with E, above, a required power density ofE,/t - 1 watt/cm2 is obtained.

49

In order to compare this power with that of the transducer, oneneeds to know the coupling efficiency of the transducer power.As described in the proposal resulting in this study, thatefficiency can be approximated from the acoustic impedances ofair and water (RA and Rw), whose ratio (rZA/Rw) is about 3x10"'.The exact expression for acoustic energy transfer, expressed asthe transfer coefficient (aT) is

aT = 4 RAR./ (RA + Rd)2

In the limit of RA >> Rw, which is true here,

aT = 4 RW/RA

which gives air/water acoustic coupling of -10.' Combining thisefficiency with a 10 watt/cm2 transducer, one estimates a powerdensity of -102 watt/cm . This clearly is much lower than therequirement of 1 watt/cm2 estimated above.

There is considerable uncertainty in this estimate for therequired power density. For example, assumption of r = 10" cmis critical to the final results. The purpose of the calculationwas to give a rough picture of nucleation energetics. It isinteresting to note that this 1 watt/cm2 power level (in thecontext of Table 3 in Section 4) is not inconsistent with the 150dB threshold power level quoted in the van Stratten patent. 6

The acoustic power level can also be considered as pressureperturbation (AP), e.g. 150 dB in air corresponds to -630 Pascal.This can be compared to the surface tension of water (E.) byassuming a particle radius (r), i.e.

AP - Es / r

This gives r-115 Am. For the droplet size range of interest (5-30 Am diameter) about an order of magnitude more power isrequired for perturbing the surface of the entire droplet, e.g. apower level of - 170dB.

The first estimate of 1 watt/cm2 considered the energetics anddynamics of perturbing a small crystallization nucleus as opposedto the whole droplet. The two estimates show the uncertaintiesin predicting the required power levels in the absence ofdetailed understanding of the microphysics. The point is thatthe threshold character of the nucleation process makes theacoustic power level critical to observation of positive effects.

50