UNIVERSITY OF MICHIGAN Engineering Research Institute Ann Arbor GENERATION OF SMOKE AT HIGH ALTITUDES U. S. War Department Contract No. W-36-039 sc-32307 (Meteorological Branch, Signal Corps) Final Progress Report for the period February 15 1948 to February 15, 1949 Department of the Armyv Project: No. 3-99-07-022 Signal Corps Project: No. 172B Submitted for the project by: C. M. Sliepcevich

UNIVERSITY OF MICHIGAN Engineering Research Institute UNIVERSITY OF MICHIGAN PROJECT PERSONNEL Both Part Time and Full Time Boothroyd, Alma C., Stenographer-Clerk Consiglio, Joseph A., M.S.(ChE), Research Assistant Dale, Stanley E., Research Assistant Dehmlow, Louis H., Research Assistant Granger, Boyd W., B.S.(EE), Research Assistant Gumprecht, Roland O., M.S.(ChE), Research Assistant Keig, John W., B.S.(NA), Research Assistant Kennedy, Verne C., B.S.(MetE), Research Assistant Kurata, Fred C., PhD, (Associate Professor of Chemical Engineering, University of Kansas, Lawrence, Kans.) Reineke, Harry, B.S.(NA), Research Assistant Sliepcevich, C.M., PhD, Assistant Professor of Chemical Engineering Tek, Rasin, B.S. (E), Research Assistant Thatcher, Charles M., B.S.(ChE), Instructor of Chemical Engineering Printed and Lithoprinted in U. S. A. University Lithoprinters, Ypsilanti, Michigan 1949

UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE OF CONTENTS Page Generation of Smoke at High Altitudes. 1 Part A: A Proposed Blethod for Quantitatively Determining the Effective Average Particle Sizes in Smokes. 3 I Experimental Equipment and Instrumentation. 3 II Experimental Procedure. 7 III Theoretical Discussion and Analysis. 12 IV Calibration of Instruments. 25 V Conclusions. 27 Part B: Comparison of Smokes Produced by FS-H20 and by Atomization of Liquids in the TOP Chamber. 28 I Experimental Equipment. 28 II Experimental Procedure. 28 III Experimental Data and Summary of Results. 30 IV Discussion of Results. 35 V Conclusions. 41 Part C: Vibrating-Type Atomizing Nozzles. 42 I Experimental Equipment. 42 II Experimental Procedure. 45 [II Experimental Data and Results. 46 IV Qualitative Observations on the Operating Characteristics of the Vibrating-Type Atomizing Nozzle. 64 V Conclusions. 65 VI Acknowledgme nt. 66

UNIVERSITY OF MICHIGAN Engineering Research Institute ILLUSTRATIONS Part A Figure No. Title Page 1 Instrument Wiring Diagram of Sboke Generation Equipment. 4 2 Lead Sulfide Cell Amplifier for Monitoring Concentrated Arc Light Source, 5 3 Diagram of Negative 1000 V Power Supply. 6 4 Pre-amplifier for Scattered Light. 6 5 Diagram of Main Amplifier. 8 6 Pre-amplifier for Transmitted Light. 9 7 Regulated 270 V Positive Power Supply. 9 8 Piping Diagram of Smoke Generation Equipment. 10 9 Schematic View of Transmitted Light and 90~ Scattered Light in Spraying Chamber. 13 10 Intensity of 90~ Scattered Light vs. Elapsed Time. 17 11 Intensity of Transmitted Light vs. Elapsed Timeo 17 12 Illuminated Fog Element at the Center of the Spraying Chairhber 18 13 Cross-Sectional View of Device to Calibrate Instrument Recording Intensity of 90~ Scattered Light. 25 Part B Fi gure No, Title Pae 14 Typical Calibration Curve for Esterline-Angus Recorder. 29 15 Comparison of Light Transmission Curves for Various Substances. 31 16 Minimum Fraction of Light Transmitted by Water and Kerosene as a Function of Volume Sprayed. 32

UNIVERSITY OF MICHIGAN Engineering Research Institute ILLUSTRATIONS (continued) Part C Figure No. Title Page 17 Mechanical Loader for Determining Initial Spring Tensions. 43 18 Typical Load-Deflection Curves for Spring No. 2 43 19 Assembly Sketch for Vibrating-Type Nozzle with Helical Spring. 44 20 The Relationship Between Spring Tension and Minimum Spraying Pressure. 47 21 Capacity versus Pressure for Nozzle No. 17, 0.032 inch Stem. 48 22 Capacity versus Pressure for Nozzle No. 17, 0.0338 inch Stem. 50 Capacity versus Pressure for Nozzle No. 17, 0.0358 inch Stem. 51 24 Comparison of Capacity versus Pressure for Nozzle No. 17, (Orifice, 0.04 inch) for Various Stem Diameters. 52 25 Comparison of Capacity versus Pressure for Nozzle No. 17, Stem, 0.0358 inch, for Various Springs. 53 26 Vibrating Characteristics of Nozzle No. 1i. 64

UNIVERSITY OF MICHIGAN Engineering Research Institute TABLES Part B Table No. Topic Page I Experimental Data. 33 II Comparison of Effective Average Particle Sizes for Water and for Kerosene. 34 III Relative Weights of Kerosene and Water Required to Produce an Equivalent Smoke Based on the Standard FS-H2O Smoke. 34 IV Fractions of Fluids Sprayed Which Settle Out Instantaneouslyo 38 V Experimental Data on Nozzle F-80NS-o.60-80~M at a Spraying Pressure of 1000 psig. 39 Part C VI Effect of Initial Spring Tension on the Minimum Spraying Pressure; Nozzle No. 17 (orifice = 0.04 inches); Various Stems and Springs. 54 VII Effect of Spring Tension and Spraying Pressure on Capacity; Nozzle No. 17, 0.032 inch Stem. 56 VIII Effect of Spring Tension and Spraying Pressure on Capacity; Nozzle No. 17, 0.0338 inch Stem (Refinished, originally 0.0358 inch stem) 58 IX Effect of Spring Stiffness, Spring Tension, and Spraying Pressure on Capacity; Nozzle No. 17, 0.0358 inch Stem. 61 X Effect of Spring Stiffness on Capacity at Various Spring Tensions and Spraying Pressures; Nozzle No. 17, 0.0356 inch Stem. 63

UNIVERSITY OF MICHIGAN Engineering Research Institute GENWERATION OF SMOKE AT HIGH ALTIT1DES Department of Chemical and Metallurgical Engineering At the request of the Meteorological Branch of the Signal Carps the University of Michigan undertook, on February 15, 1948, to study a means for generating a smoke trail on the V-2 rocket after burnout. The purpose of this smoke trail was to provide a means for studying upper atmospheric winds and turbulence. Briefly, this smoke trail was required to fulfill the following specifications: (1) A finely dispersed smoke consisting of particles in the neighborhood of 0,5 microns in diameter to be discharged continuously from the V-2 rocket in the altitude range of 100,000 to 200,000 feet in 25 seconds. (2) Sufficient smoke to produce a visible cloud at least 30 feet in diameter and 100,000 feet high. (3) The total weight of the smoke generation equipment, including the smoke producing materials, not to exceed 1000 pounds. (4) The smoke trail to be photographed from the ground and most probably at sunrise. (5) The possibility of eventually adapting this smoke generation unit to the Aerobee program, in which the "pay-load" limit would be 150 pounds. An analysis of the problem on smoke generation indicated that the present work should include the following studies: (1) The development of a method for quantitatively evaluating the light scattering properties of the various smokes. Furthermore, since for every smoke of a given particle size, a certain weight of smoke would have to be discharged in order to meet the requirements of "minimum photographability" in the Upper Atmosphere Program, it immediately became evident that any method for quantitatively evaluating the smoke should include some means for determining average particle sizes. (2) The possibility of creating a smoke trail by the atomization of certain fluids with commercially available spray nozzles. As generally known, certain fluids can be much more finely dispersed than others. In addition, based on a study of energy requirements, the method of atomization was necessarily restricted to the direct atomization of fluids under pressure. (3) The possibility of using a vibrating-type, atomizing nozzle. Previous work by the writer indicated that this type of nozzle was capable of producing not only finer, but more uniform, dispersions than any of the commercially available spray nozzles.

2 UNIVERSITY GF MICHIGAN Engineering Research Institute The above investigations were carried out for a period of one year as of 15 February 1949. At that time it was decided to terminate this particular phase of the work since it was no longer considered to be of immediate interest or concern to the Upper Atmosphere Program. A complete summary of the problems encountered in this present work, particularly with regard to experimental techniques and instrumentation, has been previously reported. For detailed descriptions, the reader is referred to the following reports which were issued by the Engineering Research Institute of the University of Michigan under the title, Atmospheric Phenomena at High Altitudes: 1. Progress Report No. 5 for the period January 15, 1948 to April 15, 1948. 2. Progress Report No. 6 for the period April 15, 1948 to July 15, 1948. 3. Progress Report No. 10, Quarterly Report No. 7 for the period July 15, 1948 to October 15, 1948o The present report will include the progress of the work for the period October 15, 1948 to February 15, 1949 (termination date), in addition to a brief summary of the entire program. Throughout this report, frequent reference will be made to the above-mentioned Progress Reports for detailed descriptions. This final report will be divided into three parts: Part A: A Proposed Method for Quantitatively Determining the Effective Average Particle Sizes in Smokes. Part B: Comparison of the Smokes Produced by FS-H20 and by the Atomization of Liquids in the TOP Chamber. Part C: Operating Characteristics of Vibratiing-Type Atomizing Nozzles.

3 UNIVERSITY OF MICHIGAN Engineering Research Institute Part A A Proposed Method for Quantitatively Determining the Effective Average article Sizes in Smokes The major problem in the present studies on smoke generation involved the development of suitable equipment for quantitatively comparing the various methods for producing smoke. For the reasons given in Progress Report No. 5, pages 33 ff, a method employing the measurement of both the amount of light transmitted through the smoke and scattered at an angle of 90~ to the incident beam was adopted. A further analysis of this method revealed that, with certain modifications and refinements in the experimental equipment and instrumentation, information could be obtained as to the average particle sizes and their distribution. (See Progress Report No. 6, pages 29 ff and page 37, and Progress Report No. 10, Quarterly Report No. 7, page 43.) Because of the great desirability for obtaining such information, major efforts were devoted to the investigation and possible development of such a method. Although much work remains to be done, considerable progress has been made in this direction. I Experimental Equipment and Instrumentation. A block diagram of the equipment for determining the effective average particle sizes in smokes is shown in Figure 1. The spraying chamber consists of a 150-gallon steel condensate drip tank (30 inches ID and 48 inches high). A 300-watt, 320-candle-power, Western Union, Type K300 AC concentrated arc lamp and power supply unit serve as a light source. (Progress Report No. 10 Quarterly Report No. 7, pages 27 ff). This arc lamp has a cathode area of 0.012 sq. inches and, consequently, furnishes practically a point source of light. Modulation of the beam is accomplished by use of a perforated aluminum disk having ten 5/8 inch holes equally spaced on a diameter of 4 inches. It is driven by an 1800 RPM, 2-watt, General Electric, synchronous motor; thus, a major frequency component of 300 cycles per second is generated. The modulated light beea passes through a collimating lens system (f/3.5 Argus Viewer Lens - 3" focal length) to give an essentially parallel beam. The intensity of the light source is continuously measured or monitored during a test run by means of the circuit shown in Figure 2. The parallel beam from the light source passes through a piece of plain glass placed at an angle of 45~ with the beam. A small portion of this beam is reflected downward onto a small Cetron lead sulfide photocell. This signal is amplified and recorded on a strip chart recorder (Leeds & Northrup Type G Speedomax). Monochromatic light is obtained by the use of a band interferencetype filter manufactured by Baird Associates, Inc. According to their specifications, this filter has a peak transmission at 4700 A +~ 50 A and a band width of 100-150 A at one-half maximum transmission. In effect, the filter operates as a fixed separation Fabry-Perot interferometero The use of blue light is dictated by the spectral response curve of the 931A Photomultiplier tube, which does not respond to light of wave lengths greater than 7000 A.

110 V AC CONSTANT VOLTAGE 500 TOSOLA TRANSFORMER n VA ===~~TOIIOV AC ----- VA CONSTANT VOLTAGE 120 VA SOLA TRANSFORMER 450 WATT POWER SUPPLY 8 WIRE SHIELDED.p 1800 RPM / \, CABLE:1 SYNCHRONOUS SPRAYING WRI 43 MOTOR CHAMBER SHILDED rl ~~~~~~~~~~~~~~~~~~~~~~~~SHIELDED SI CABLE CABLE MAIN REGULATED 0 ~ ^ r"' [^ L' =: i,,MriinM AMPLIFIER POWER J3 fO 1'IV JUNCTION SUPP t~ ~~ AC BoxSUPL 0~i *g I ^\ TRANSMITTED 09 300 WATT CONCENTRATED LIGHT DETECTOR AqC LIGHT 0 0. 4)Q 2 WIRE SHIELDED CABLES ~ 4 900 SCATTERED LIGHT CO VT DETECTOR 490 SCATTERED TRANSMITTED ff [___________________________, LIGHT CHART LIGHT CHART 6- WI.RE SHIE^~LDED -- ~RECORDER RECORDER \6 WIRE SHIELDED CABLE Pb S CELL --- A OR ----— ___ INSTRUMENT WIRING DIAGRAM OF SMOKE MONITORING GENERATION EQUIPMENT LIGHT CHART RECORDER Figure lo 10

5 UNIVERSITY OF MICHIGAN Engineering Research Institute -+140 +140 20K IN 34.O5A5fdfd _ - -[_....L__ —__ 1959 5.05 fd 955 Pb x CELL Meg o~l map. TO STRIP CHART ^'Q CELL 0' a.1 0" meg' Al. I' —~' U d ^RECORDER +J_ 8 Ufd 25 Ufd ~. 22.5v 330n - - 300 Note xx to 6.3 v battery |14 3v LEAD SULPHIDE CELL AMPLIFIER FOR MONITORING CONCENTRATED ARC LIGHT SOURCE Figure 2. The transmitted and scattered light detectors contain suitable lens-pinhole systems, of the type described in Progress Report No. 10, Quarterly Report No. 7, pages 24 ff. This arrangement limits the transmitted and scattered light tolerances to less than plus or minus one degree. In addition, both the transmitted and scattered light detectors contain the necessary photo tubes and preamplifiers. From empirical data obtained during some of the preliminary runs, it was estimated that the amount of illumination that would intercept the effective cathode area of a phototube in the scattered-light detector was in the order of 4 x 10-7 lumens. (See Progress Report No. 10, Quarterly Report No. 7, page 32.) Because of this relatively low level of illumination, the use of a photo-multiplier type of vacuum phototube and a band-pass amplifier, capable of amplifying a single alternating component (modulated at 300 cycles), was necessitated in the case of the scattered-light detector. Since in such a tube there are nine small tubes or dynodes, each so designed to operate at maximum efficiency at 100 volts per dynode, it was necessary to construct a 1000-volt power supply. (Compare with Progress Report No. 10, Quarterly Report No. 7, pages 28 ff.) This supply was designed as a constant-voltage, current-compensated device which remains stable despite any changes in the load current resulting from variations in the incident light beam. A circuit diagram for this power supply is shown in Figure 3. The phototube, its high impedance circuits, and a cathode-follower type of impedance transformer, as shown in Figure 4, were enclosed in a separate preamplifier chassis. After preamplification, the signal passes through a doubly-shielded cable to the main amplifier as represented by the

6 UNIVERSITY OF MICHIGAN Engineering Research Institute T I AMP CATHODE 6F6 x o 2X2'WATT NEON 6F6G 100 K 100 K -100 5o Moo Rme DIAGRAM OF NEGATIVE. OO v KP SUPPLY Figure 3..5 n\2,m -1000 a 210 v B+ ~.5 meg' /:~0 K 47K1 K: Of i XX) L.. OUTPUT - KOOOv PRE-AMPLIFER FOR SCATTERED LIGHT Figure 4o

7 UNIVERSITY CF MICHIGAN Engineering Research Institute circuit diagram of Figure 6. The signal first enters an attenuator calibrated in fixed 3-db steps. From this voltage divider the signal passes through a narrow band pass filter which is composed of two, resistance-capacitance, parallel T circuits used in conjunction with their respective degenerative pentode amplifiers. The net effect is a circuit that is degenerative at all frequencies except 300 cycles. (Compare with Progress Report No. 10, Quarterly Report No. 7, pages 37 ff.) The 300 cycle signal is next rectified, filtered, and applied as an average DC value to a degenerative cathode follower amplifier which was designed so as to match the 25 ohm impedance of the Leeds and Northrup Micromax Recorder (Model S). A similar system was employed for the amplification of the transmitted light signal except that an ordinary 929 vacuum phototube was sufficient for the higher light levels obtained by transmitted light. For this same reason, no filter stage was required. Figure 6 gives the circuit diagram of the preamplifier for transmitted light. Because of the extreme circuit stability desired, (Progress Report No. 5, pages 40 ff) the 270-volt plate voltage supply for the main amplifier was designed as an electronically regulated type with constant voltage, voltagecompensated supply as is shown in Figure 7. Both the 1000-volt negative supply (Figure 3) and 270-volt positive supply (Figure 7) were mounted on the same chassis. (Progress Report No. 10, Quarterly Report No. 7, page 28) The transmitted and scattered light signals from the main amplifier are continuously recorded by two Leeds and Northrup Micromax Recorders (Model S) -- see Progress Report No. 5, page 35. II Experimental Procedure. The material in this section describes in detail the procedure followed in conducting a fog test with the vacuum chamber and its auxiliary equipment. In order to follow the description, the reader is referred to Figure 8 which is a diagrammatic representation of the fog chamber with its auxiliary equipment and piping layouts To evacuate the chamber, all valves except Nos. 14, 15 and 19 are closed and the vacuum pumps are turned on and allowed to run until the required pressure is reached. The closed tube manometer is provided to measure chamber pressures in the range 0-160 mm Hg, and the McLeod gage is used for pressures below 15 mm Hg. When the desired vacuum has been reached, the pumps are turned off and valves Nos. 14 and 15 closed. The liquid to be atomized is introduced into the spraying solution reservoir with valves Nos. 1 and 10 open. When the reservoir is full, these valves are closed. Then, by opening valves 11 and 20, air pressure is applied to the top of the liquid in the reservoir. The delivery lines are flushed out and filled with the liquid to be sprayed by opening valves 2 end 9 for a short time, after which valve 9 is closedo With valves Nos. 3, 4, 8, and 9 closed, and 2, 5, and 7 open, the positive displacement chamber is charged with the liquid to be atomized. The amount of liquid to be sprayed is predetermined

270 vB+ DIAGRAM OF MAIN AMPLIFIER 22 K _ -6AL5 2 -2AU7.osfd.05,ufd i-12AU7 2 -12AU.O05ufd 05d fd INPUT FROM LIGHT PRE-AMP- 5000 5000.4 | LIFIER. 100 4 1r4 4) 4 250K I Q 25 30K 300 300 8a.fd 3 | a:: ~* ^^pifd -; K + K 3 0'8K fdK HM 0ri< Tf-I~~ ~ | | S300 RECORDER r,150 9: 1050 n 1000 REOORDER INPUT FROM SfATT- 1-12AU7.05pfd I meg 2 ERED LIGHT PRE- 0 I me meg R I AMPLIFIER P,.5 fd..6ufd 0 M |~22K f ^n ^^ Y.^ R, R2 RI 150 330 R1 2 RI 47K 150K K K C2 C2 270 V B+ Figure 5. cD

9 UNIVERSITY OF MICHIGAN Engineering Research Institute 270 v B+ 20 Mg. 20 K 8p tfd.O 5fd/ 6J4 929 ( J Mem60 meg iKOO n OUTPUT PRE- AMPLIFIER FOR TRANSMITTED LIGHT Figure 6. 6 54 REGULATED 70 V P27TIVE POWER SUPPLY oS~~ ~~10 HEN IK n. 0 K 100 K 40 )Jf._. 0 K --— =OOKZ (^r — J50K'SH 5V4 + 100 K_ I10 ~, 5651 SOLA 25 K REGULATED 270 V POSITIVE POWER SUPPLY Figure 7.

INTAKE ATMOSPHERE TO COMPRESSED I^~ INTAKE __~( () ^ OUTLET AI - 10OOO psi AIR s 0o-?o T I' — L-, —-. —--------- ---— DRAIN - SPRAYING I SOLUTION SPRAY (OT RESERVOIR NOZZLE C ~~_-~~~~~_ -—:^^^^~~~~~~~ ~ ~ ~ ~ Jl^^^^-~ IDRYING CLOSED TUBE - -:',~~~~~ ii~ /~'~,~,~\^~.MANOMETER 4 I) FILTER _r U -- E~TMOS ~~~~~~3 I —i —I ni-ic.M: _i^ -_^ — (Q~~~~ (1~ OUTLET' _ 3 @^ I|'''-~:=~. ~ |INTERRUP- SPRAYING CHAMBER. EATIN a5 gI — TED LIGHT C CI DRYING CTLEOD IGT H0 S |~ - PISTO SOURCE U GGE TO POWER TO AMPLIFIER I C),r p_ OIL IN LE _~ _ PUMPS P|i TO AMPLIFIERPING DIAGRAM ~-: ~z ~=^ ~'=~RA SMOKE GENERATION EP l.,,,,_ TR IN T!igr I Im 2;:::1 TO DRAIN ( ATMOSPHERE VRE 3 HASE Figure 8. o Fi g u r e ~~~~~~~~~~~~~8.

11 UNIVERSITY OF MICHIGAN Engineering Research Institute by screwing the threaded stem on top of the positive displacement chamber to the proper level indicated on a calibrated scale which is mounted beside the stem. The floating piston moves upward as the positive displacement chamber is being charged and strikes the threaded stem with an audible "thud", at Riich time valve No. 2 is closed. The pressure in the reservoir is released by closing valve No. 20 and venting through valve No. 10. Prior to the time that the above preparations are being made, the light source, power supplies, amplifiers and strip chart recorders are turned on and allowed to warm up. The "zeroing' rheostats in the amplifiers are adjusted so that the strip chart recorders are zeroed at convenient locations on their charts. When the strip chart recorders indicate that steady conditions prevail, the system is then ready for fog testing. With valves No. 5 and 7 open, the oil pump is started. Pressure is built up in the positive displacement chamber by partially closing valve No. 7, thus throttling the oil recycle to the oil reservoir. When the gage indicates that the desired pressure has been reached, valve No. 8 is opened and spray commences. A stop watch is used to record the length of time required to complete the spray. During spray the pressure remains very constantbut at the end of the spray a sudden decrease in pressure is indicated on the gage,at which time the stop-watch is stopped. The spraying pressure, fluid sprayed, temperature of the fluid, quantity sprayed, time of spray, type of nozzle, fog chamber pressure, time of day, test number, and date are recorded. The test number is also recorded on each of the strip charts recording the various light intensities. After sufficient time has elapsed and the strip chart recorders have drawn their decay curves (see Progress Report No. 5, page 45), the vacuum in the fog chamber is broken by venting through valve No. 16. The vacuum in the closed tube manometer and McLeod gage is broken by venting through valve No. 18 qo that only dry air is allowed to enter this part of the system. The fog eoamber is then flushed out with compressed air by opening valve No. 13, 11, and 12. If a corrosive fluid such as FS has been atomized in the fog chamber, it is necessary to wash the inside of the tank with water by opening valves 23 and 24. After the tank has been thoroughly washed, valve No. 23 is closed and the inside of the chamber dried by blowing in hot dry air with valves 21, 22, end 24 open. In actual operation, the floating piston in the positive displacement chamber has occasionally become stuck at the bottom of the chamber. When this happens, valves Nos. 2, 3, 5, 8, and 9 are closed and valves Nos. 4 6, and 7 are opened. The oil pump is started and valve 7 throttled down, causing pressure to build up under the piston and thereby dislodging it. After this operation it is necessary to flush out the positive displacement chamber and spray lines so that the fluid sprayed on the following test will not be contaminated with oilo

12 UNIVERSITY OF MICHIGAN Engineering Research Institute III Theoretical Discussion and Analysis, The previous progress reports from this laboratory have described in detail the equipment which has been installed for the purpose of quantitatively evaluating the light scattering properties of various artificially generated fogs. By making use of the data obtained in testing the fogs, the possibility of calculating the effective average fog droplet size presented itself. Since the data obtained consist of simultaneous independent measurements of the intensity of a parallel bean of light after it has passed through the fog and the intensity of light scattered at an angle of 90~ to the bean, the possibility of applying the Mie* equations for light scattering by spherical particles was considered. The Mie equations define, not only the magnitude of the total light scattered by a spherical particle illuminated by a parallel bean, but also the angular distribution of the scattered light around the particle. The chief difficulty in directly applying the Mie equations to the data obtained is the fact that the Mie equations are valid only in the case of infinitely thin fogs and do not cover secondary and higher orders of scattered light, i.e., light which undergoes scattering as a result of striking two or more particles in succession. In a recent paper, however, Theissing** has extended the Mie theory of light scattering and has shown that secondary and higher orders of scattered light bear a quantitative relationship to the primary scattered light both in magnitude and angular distribution for fogs of finite density. Therefore, by means of the Mie and Theissing equations, it has become possible to calculate the effective average particle size of a fog existing at any chosen instant of time after its generation in the testing chamber in this laboratory. The following section will deal with the development of the data obtained in this laboratory into quantities which can be substituted into Theissing's equations for obtaining the effective average particle size. The nomenclature to be used throughout is defined in Table I. TABLE I Nomenclature I = intensity of the parallel beam of monochromatic light before entering the fog (lumens) I a intensity of the parallel beam of monochromatic light after passing through the fog (lumens) 1 a intensity of the parallel beam just as it reaches the fog element in the center of the fog chamber (lumens). (See Figure 9) Sgo0 intensity of the light scattered at an angle of 90~ to the parallel beam by the fog element at the center of the chamber. (lumens per steradim * G. Mie, Ann d. Phys. 25, 377 (1908) ** H. Theissing, "Application of Light Scattering I", Evans Signal Laboratory, Tech. Memo M-1149, Belmar, N. J., 15 October 1948

13 UNIVERSITY CF MICHIGAN Engineering Research Institute SCHEMATIC VIEW OF TRANSMITTED LIGHT AND 90~ SCATTERED LIGHT IN THE SPRAYING CHAMBER. _ -- d, - d, d4 | I \ I | di d d,= 17.75 in. d2= 19.0 in. I=35.5 in. d3= 1.06 in. \ d4= 1.12 in. S90 Figure 9.

14 UNIVERSITY CF MICHIGAN Engineering Research Institute TABLE I (cont'd) SO = intensity of the light scattered at an angle of 90~ to the parallel beam by the fog element as it leaves the fog chamber at the side window and is recorded by the 90~ scattered-light measuring instrument. t S9 = 90 = intensity of the light scattered at 90~ to the parallel beam per 1. unit of light entering the fog element. (lumens per steradian per lumen) q = scattering cross sectional area per droplet (cm2) n = number of droplets per unit of volume (cm-3) J - = length of the path of the parallel beam through the fog chamber (an) o0 = A —-- D = droplet diameter (microns),A. wave length of monochromatic light in the beam (microns) M1 M2 M3 etc. = Theissing's magnitude functions of primary, secondary, etc. scattered light. They represent the total number of lumens of primary, secondary, etc. scattered light emerging from the illuminated fog element in all directions per lumen of light striking the fog element in the parallel beam. fl(90), f2(90), f3(90) etc. = Theissing's angular distribution functions of the various orders of scattered light emerging from the fog element at an angle of 90~ to the parallel beam. They are expressed in terms of lumens per steradian per lumen of primary, secondary, etc. light scattered in all directions~ Theissing's generalized equation for light scattering by spherical droplets appears as follows: S90 = Mlf (90) + M2f2(90) + + + kfk(90). Therefore, the first task is to calculate the values of S90, M1 M2 M3 etc. from the experimental data obtained. Theissing has shown that the angular distribution functions, fl(90), f2(90), etc. are definite known functions of o(, which is the ratio of particle circumference to wave length of light in the beam. Therefore, the problem of calculating the effective average particle size becomes one of choosing the proper value of o< such that, when the angular di stribution functions thus obtained are substituted into the general equation along with the values of Sqos 1, 2A2, etc. which are caltculated from the experimental data, the equation will be satisfied. After the value of O( is obtained, the effective average particle diameter is easily obtained since o<A D> ^ T

15 UNIVERSITY OF MICHIGAN Engineering Research Institute In general, the decrease in light intensity of a beam of light in a fog may be represented as follows I _ - = qn dx where dx is a differential unit of distance along the path of the beam. This expression integrates into the form of the familiar equation I _ qnl I = e (1)* and will be used frequently in the following derivations. Refer now to Figure 9 and consider the small element of fog in the center of the fog chamber, which is "seen" by the instrument recording the intensity of light scattered at an angle of 90~ to the incident beam. For the time being, assume this fog element to be bounded by a small cylinder of diameter d4 and length d3. Until measurements can be made which will indicate the degree of true absorption (i.e., to Joule heat) of the fog droplets under consideration, this factor will be neglected in the following derivations; therefore, losses of light intensity by the beam travelling through the fog will be assumed to be due to scattering alone. Since the fog element under consideration is located at the geometric center of the fog chamber, by Equation (1) I1 e qn (dl 2 ) (2) Io Also S0 = e qn dl SI (31 From equation (1) - qn = 1 in ) Equations (2) and (3) then become by substitution: 2dl d3 - d I = Io ( o); g = S0 ( I 10 0Io * Note: This equation neglects true absorption and attributes the total decrease in light intensity to scattering, alone. It also assumes a uniform distribution of particles along the entire path of the beam. See Progress Report No. 6, pages 29 ff.

16 UNIVERSITY OF MICHIGAN Engineering Research Institute Therefore, d SgO= S0 O = S90 (10/ I1 2al - dd3 I90 -o ~o\ IJ which simplifies to: d3 S90 S90 I I \Io Since the scattered light recording instrument is calibrated in foot candles, it is necessary to convert its readings into lumens per steradian for insertion as S~o in equation (4) above. One square foot of surface on a sphere of one foot radius subtends an angle of one steradian; therefore, since the lens is located 19" from the fog element and one 2square foot of surface on a sphere of radius 19* subtends a solid angle of12 steradian, the foot candle readings must be divided by (~.9) in order to express S90 in terms of lumens per steradian. Equation (4) then becomes: I Io S9go= (sI ) ) x (2.51) (5) Equation (5) may then be evaluated by using the values of Sgo in foot candles and I in lumens as read from the ordinates of the curves shown in Figures 10 and 11 respectively, at any chosen instant after spraying has been completed. Theissing has shown that the various quantity functions M1, M2, M3, etc. of the various orders of scattered light are related to the fog density as follows: M1 qn x e- qnx (6) 2 = (qnx)2 e qnx M2 t -"i2 — e (7) M)k= (qnx)k qnx (8) Where x in this case refers to te length of the path of the incident beam through the fog element under consideration.

17 UNIVERSITY OF MICHIGAN Engineering Research Institute INTENSITY OF 90' SCATTERED LIGHT VS ELAPSED TIME 12 Z MATERIAL SPRAYED- 100 CC KEROSENE (U.M STOREHOUSE) 0 FOG CHAMBER PRESSURE — ATMOSPHERIC PRESSURE lCt,&~~~~~~~~ NOZZLE KOPP H 90o 0.75 _t-l~~o~~ ~~SPRAYI NO PRESSURE F 11 I I I I 1 1000 psi 0 0 TEST L-I z B | O 00 psl O — TEST L-2 E%~ - 600 psi X-* —X TEST L-3 ~Qt~~ L\U~'"^<S~~ _ l | DATA OBTAINED FEB. 3,1949 I- X. XX'-X'-x — X --— X —43 0r 0 20 40 60 80 100 120 140 160 180 TIME FROM END OF SPRAY (MINUTES ) Figure 10. |r-^ I |INTENSITY OF TRANSMITTED LIGHT vs ELAPSED TIME i' 0 3 z z I- I/ IATERIAL SPRAYED -100 CC KEROSENE (U.M.STEHO / S FOC CHAMER PRESSURE - ATMOSPHERIC PRESSURE - t/ NOZZLE KOPP H 0 0.75 I- ^ A SPRAYING PRESSURE _ / ____ __________ ________ I 1000 psi Oo TEST L-I, t,0401 LUMENS Z z 8600psi tm —- TEST L-2, I.-a66 LUMENS //00 osil --- I Ie TEST L- It.0:73 LUMENS DATA OBTAINED FEB.3,I)4 0 2 0) 4o.o 0 0 oo 12~ 0 ) 40 10 iB TIME FROM END OF SPRAY (MITES) Figure 11.

18 UNIVERSITY OF MICHIGAN Engineering Research Institute It was previously assumed that the fog element shown in the center of the chamber in Figure 9 is bounded by a cylinder of length d3 and diameter d4; however, this assumption is not actually true since the fog element is in reality bounded by the intersection of two cylinders. The diameter of the cylinder containing the incident beam is d4 and the diameter of the cylinder projected back toward the 90~ window is d3. These diameters were measured and found to be equal to 2.84 cm and 2.69 cm respectively. The immediate problem then is to evaluate the total light scattered by the fog droplets bounded by the intersection of these two cylinders so that the fog element may be considered to be bounded by an equivalent cylinder of diameter 2.84 cm and length x. The value of x could then be used in equations (6), (7) and (8). Referring to Figure 12, let the equation of the cylinder of the light beam be represented as: y2 + z2 (1.42)2 and the cylinder intersecting at right angles: x2 + z2 =(1.345)2 LIGHT BEAM IN FOG CHAMBER /, z dz 2.69 cm. REGION WITHIN WHICH LIGHT SCATTERED BY FOG DROPLETS IS RECEIVED AND RECORDED BY 90~ SCATTERED LIGHT MEASURING INSTRUMENT. ILLUMINATED FOG ELEMENT AT THE CENTER OF THE SPRAYING CHAMBER. Figure 12.

19 UNIVERSITY OF MICHIGAN Engineering Research Institute Consider a horizontal differential slice of the fog element of width 2y, length 2x, thickness dz, and located at a height z above the center of the fog element. Again, let I1 be the beam intensity at the left edge of the fog element, where the light is travelling from left to right. The beam intensity at the left edge of the differential slice will then be: Ix = I1 e- qn(1.345 - x) in lumens The number of lunens entering the slice will be; Il e- qn(1.345 - x) \ -I ----------- (2y dz) \ 7T (1.42)2 The number of lumens leaving the slice in the parallel beam will be: I1 e- qn(l.345 + x)) (142)2 / ( 2y dz ) By difference, the number of lumens scattered by the fog particles in the differential slice will be: ds t I12y e qn(l.345 - x) qn(l.345 + x) /d (9) Tr (1.42)2 J From equation (1), -qn was found to be equal to ln n 92 ln'9 = 90.2 i By substituting this value of (-qn) as well as the values of x and y in terms of z back into equation (9) one obtains: 90.2 90.2 90.2 dz 1.345 1.345 (1.422 z / d' (10) dst iT 2~ 1042(1.42)2 J

20 UNIVERSITY OF MICHIGAN Engineering Research Institute If this amount of scattered light is equated to the amount of light scattered by an equivalent fog element, cylindrical in shape, at the center of the chamber whose diameter is 2.84 cm and length x, then the qnx value required to solve for M1, M2, etc. in equations (6), (7) and (8) may be readily determined. In calculating the amount of light scattered by the equivalent cylindrical fog element, a very close approximation may be obtained by assuming Ix = I1 then: t = I1 (1 - e qnx) (11) By equating equation (11) to (10) 1.345345 90*.2 (1"'-T O~r -r )' / e' -z7 /d (1 -qnx lr-(1.4 ),|1.q42)2 1z2 (.4d From which it is found that x, the length of the equivalent fog element, has a constant value of 2.17 cm, independent of the value of ( I \Io Therefore, 2 2.17 (O q- 90.2 CPI / (12) This value of qnx may now be substituted into equations (6), (7), and (8) with the following result: 2.17 90.2 1 90.2 IO n (13) 2.17 2 = (9 2.17 In(I (14 }k - -)'-',,90.2 _ T (15)

21 UNIVERSITY OF MICHIGAN Engineering Research Institute Referring back to equation (5) and substituting the value 2.17 for d3 and 90.2 for,, one obtains: 390-(^ fr)^ ( )(16) S90 (S90) 180'4 2.51 (16) The procedure used in calculating the effective average particle size from the experimental data is as follows: At some chosen instant after completion of spray, read S90, I, and Io from the charts shown in Figure 10 and 11. Calculate Sg0 by equation (16) Calculate MIlM2,etc. by equations (13), (14) and (15) By trial and error find an o< such that when the f1(90), f2(90), etc., which are obtained from Theissing's relationships between K( and the angular distribution functions, are substituted into Theissing's general light scattering equation, it will be satisfiedo For example, consider the fog particles suspended in the fog chamber 120 minutes after completion of spraying for test L-l From the charts shown in Figures 10 and 11: S0 - 3.75 x 106 foot candles I =.0317 lumen Io =.0401 lumen (7LI = 0.79 By equation (16) 3.75 x 10'6.0121 S90.17 x 10' (.79) xe.51 = 2.97 x 10' By equation (13) M1 902 ln (1.265) X (.79) 5.63024 x In this case, M2, Mr3, etc., are negligibly asmall and may be dropped. Then S90 = M1 fl (90)

22 UNIVERSITY OF MICHIGAN Engineering Research Institute From which fl (90) is calculated to be equal to 5.28 x 10-2 To solve for (i, one now needs a chart of fl(90) as a function of o( for fog droplets of index of refraction equal to that of kerosene. To date, this chart has not been calculated in this laboratory. Therefore, for the purpose of illustration, Theissing's table of fl(90) ve. o( for droplets whose index of refraction is 1.25 will be used. From this table it is seen that fl(90) =.0528 at CO = 0.9 therefore 0o X (0.>9) (.47) D = 7 3 = 0.14 micron It is realized that the above calculations are only approximations. Tables of f (90) as a function of c. must be prepared and extended into regions of higher o values for liquids of various indices of refraction. Farther refinements must be incorporated into equation (16) for S9o since it is found that at higher fog densities fl(90), f2(90), etc. are calculated to be equal to quantities greater than 1/41". The previous calculations have not discussed the accuracy with which the values S90, I, and Io are obtained with the equipment in this laboratory. Perhaps the first question to arise is relative to the effect of the chember windows on the values of transmitted and scattered light measurements. The light losses through the chamber windows have been cancelled out since during the calibration of each instrument, light of various known intensities was allowed to pass through a chamber window before reaching the recording instrument. Thus, the recording instruments are calibrated in terms of the light intensities at the inside of the chamber windows and not in terms of the aunt of light which actually emerges from the chamber through the windows. A more detailed discussion of calibration methods appears on page 25 Fogging of the observation windows was eliminated in tests L-l, L-2, and L-3 (which are plotted in Figures 10 and 11) by rubbing a thin film of soap on the inner surfaces of the windows prior to instrument calibration and the test runs. Another important consideration is the percentage of light scattered in the forward direction which is recorded as "transmitted* light. The lenspinhole system employed with the transmitted light measuring system is composed of two lenses of diameter, 7.5 cm, and focal length, 17.5 cm, placed close together so that the focal length of the combination is 8.75 cm. A pinhole 1.6 mm in diameter is placed at the focus behind the lenses. Considering the forward scattered light from a droplet near the light source side of the fog chamber and 92 cm from the lens in the transmitted light detector (see Figure 9), by the lens law, 1/p + l/q = l/f, the image distance, q, is calculated to be 9.67 cm. behind the lens. Since this image appears behind the pin hole, the pin hole will impose a restriction on the amount of light which would compose the image. In so far as the scattered light from this particular droplet is concerned,

23 UNIVERSITY OF IMICHIGAN Engineering Research Institute the pin hole has the effect of reducing the light gathering power of the lens to one of diameter.16 x 8.75 = 1.68 ca. The general expression for the effective lens diameter turns out to be.16 p d = 8.75 where p is the distance from the droplet to the lens In centimeters. The solid angle within which light scattered in the forward direction will be recorded along with the transmitted light is therefore -fT d2 d2 - x 41 it = 4 X 4Tp2 4 p2 substituting the value of d in this expression gives the following i (.16)2 p2 - (.16)2 4 (8.75)2 p2 - 4 (8.752 = 2.63 x 10 4 Steradian Therefore, only that light scattered by any particle in the forward direction which is included within a solid angle of 2.63 x 10-4 steradian will be recorded with the transmitted light. In order to estimate the percentage of the light which is recorded as "transmitted" which is actually forward scattered light, it is obvious that the greatest errors in measurement would occur when the droplets have an extreme Mie angular distribution function, i.e., when scattering in the forward direction is much greater than any other direction. Therefore, for the purpose of estimation, assume that the fl (forward direction) is a thousand times greater than fl(90), or approximately 7 umens- per lumen scattered. Then, for each lumen scattered in all steradian directions, (7) (2.63 x 10l4) = 1.84 x 10-3 lumens will be scattered in the proper direction and will be recorded by the transmitted light recorder. In differential form, consider those particles at a distance X in the fog from the light source. The beam intensity at this point would be Ioe" qnx. The total light scattered in all directions by a differential element of fog of length dx is: dSt = I e-qnx qn dx

24 UNIVERSIT OF MICHIGAN Engineering Research Institute The anount of light scattered in the previously specified forward direction is: dSf = (1.84 x 10)-3 Io e- qn dx This forward scattered light is also reduced, however, in travelling the distance, a - x, to reach the lens of the transmitted light recorder. Therefore, the amount of forward scattered light which would actually reach the lens is: dSp = (1.84 x 10-3) I e- qnx qn. e qn-x) dx = (1.84 x 10-3) Io e q qn dx The total amount of forward scattered light from particles all along the beam which would be recorded is: SF = (1.84 x 10-3) qn Io e- qn since I qnf qn n= iln (I and I e I S_ = (1.84 x 10-3) I in (L \~7 F. (1.84 x 10-3) n (Io (17) Thus, in order for the forward scattered light recorded by the transmitted light recorder to amunt to as much as one per cent of the transmitted light recorded, the fog must be so dense that I =.0043 From the above calculations it can be readily seen that the amount of light scattered in the forward direction, which is recorded with the transmitted light, is a negligible quantity.

25 UNIVERSITY OF MICHIGAN Engineering Research Institute IV Calibration of Instruments. In order to calibrate the scattered light recording instrument, it was found necessary to expose it to known light intensities of the order of magnitude of 10-6 foot candles. Since no commercial photometer was available which could directly measure light intensities this low, it was necessary to construct a device by means of which light of known high intensities could be reduced by known reduction factors to light of low intensities. A crosssectional view of this device appears in Figure 13. It consists of a long wooden light-tight box 6" x 6" x 10' intersected on one end by a short box 6" x 6" x 2' at an angle of 45~ with the long box as shown in the diagram. The walls are j inch plywood and the interior and exterior are painted flat black. The parallel beam of light enters the short box from the light source through two holes one inch in diameter and centered on the axis of the short box. The beam strikes the ground glass screen so that the center of the spot lies on the axis of the long box. Part of the light beam is transmitted through the ground glass and is recorded by the photometer cell placed in a fixed position at "B". The diffused light emanating from the front of the ground glass travels down the long box where its intensity decreases with the square of the distance from the ground glass. Finally, it passes through one of the observation windows used on the fog chamber, through the Baird Interference type light filter, and then into the lens of the 90~ scattered light detector. PHOTOMETER CELL POSITI'ON A\ XFOG CHAMBER WINDOW / / \ STOP I /STOP 2 \ BAIRD INTERFERENCE y\.Hi~~~~~~~~~~~~~~ \ / / |\~ / ~FILTER (4700 AW) ROUND iLA \ 4 /90 SCATTERED GROUND GLASS \ SRE LIGHT DETECTOR SCREEN \ X LIGHT SOURCE d 122T GROSS-SECTIONAL VIEW OF DEVICE TO CALIBRATE INSTRUMENT RECORDING INTENSITY OF 90 SCATTERED LIGHT. Figure 13.

26 UNIVERSITY OF MICHIGAN Engineering Research Institute Stops 1 and 2 shown on the diagram are pieces of l-inch plywood through which holes 2 inches in diameter have been bored and so located that the holes are centered on the axis of the long box. These stops are placed approximately one third of the length of the box from each end. The effect of the stops is to prevent the light detector from "seeing" any illuminated surface of any wall within the box. Thus, tie only light which the detector "sees" is received directly from the ground glass screen. In order to obtain the known reduction factor referred to above, i.e., a factor by wich "B" photometer readings could be converted to light intensities at the lens of the light detector, the cell of the photometer was inserted in the long box at position'. at an accurately measured distance, d, from the ground glass screen. By multiplying "A" readings by d2 (inverse (122 P)2 square law), the white light intensity at the lens of the detector was calculated. In order to check the effects of wall reflections on "A" readings, this process was repeated at various distances, d, and the intensity at the lens calculated in each case. It was found that in order to get constant values of the calculated light intensity at the lens, it was necessary to obtain very black walls in the long box between stop 1 and the ground glass screen. This effect was accomplished by repainting the walls and dusting them with carbon black (soot) while the paint was still sticky. The calculated white light intensity at the lens was then divided by the "B" reading to obtain the white light reduction factor. The reduction factor of the interference filter was obtained by comparing strip chart recorder readings, which were taken with the filter in place, (as indicated in the diagram) and with the filter removedo In calculating the filter factor it was necessary to make use of the accurately calibrated attenuator in the main amplifier. For example, it was found that to produce a given deflection on the micromax while using one attenuator setting,always required a signal which is a constant fraction of another signal on another attenuator setting. Thus, if signal #1 on attenuator setting #1 produces the same micromax deflection as signal #2 on attenuator setting #2, then signal #1 is a constant fraction of signal #2, regardless of the value of the micromax deflection. It was necessary to make use of the attenuator in this manner since the main amplifier doesn't have a perfectly linear response. The filter factor could not be accurately obtained by merely determining the ratio of the micromax readings obtained with the filter inserted before the lens and with the filter removed. The white light reduction factor was then multiplied by the filter reduction factor to obtain the overall reduction factor. Then, with the photometer cell removed from position A and placed in position at B, the light intensity striking the ground glass screen was varied by inserting evenly exposed photographic films of various densities between the light source and the short box. For each light intensity produced, the "B" reading was multiplied by the overall reduction factor to obtain the light intensity at the lens of the scattered light detector. The latter intensities were plotted against corresponding micromax deflections to obtain the calibration curve.

27 UNIVERSITY OF MICHIGAN Engineering Research Institute The fog chamber window was placed before the lens of the 90~ scattered light detector so that its effect on the scattered light from the fog element in the center of the fog chamber could be cancelled out. The light source used in calibrating the instruments was the same 300 cycle interrupted light source that is used to project the parallel beam through the fog in the fog chamber. The transmitted light detector was calibrated in place by inserting the photometer cell in the fog chamber through the observation port and directly reading the various beam strengths. Micromax deflections were then plotted against the corresponding beam strengths to obtain the calibration curve. Here again, window effects were cancelled out. Both scattered light and transmitted light measuring instruments were calibrated using the same photometer so that any error in the absolute value of light intensities recorded would be cancelled out due to the fact that the equations developed deal with ratios of light intensities. The width of the fog element "seen" by the 90~ scattered light recorder was measured by projecting a beam approximately 1/4 inch in diameter through the long box. The beam intersects the axis of the box at 90~ and at a point just 19 inches fror the lens of the recorder. A stiff straight wire was then inserted through a slot in the top of the box at this same point so that it intersected the beam at 90~. This illuminated wire was then moved transversely across the box and the corresponding micromax readings were noted. Since the lens is focussed so that the image of an object 19 inches from the lens falls exactly on a 9/64 inch hole, 2- inches behind the lens, the system has a sharp response and cut-off when the wire is moved along the beam. V Conclusions. From the results of the tests made to date, it appears that the experimental equipment, technique, and method for determining the effective average particle sizes of various smokes has been developed.

28 UNIVERSITY OF MICHIGAN Engineering Research Institute Part B Comparison of Smokes Produced by FS-H20 and by Atomization of Liquids in the TOP Chamber The purpose of this phase of the investigation was to compare the relative weights of kerosene and of water, both atomized under pressure, with the quantity of FS-H20 which are required to produce a smoke of a given density in a 1000-cubic foot chamber. Quantitative determinations were made by means of measurements on the light transmitted through the various smokes. I Experimental Equipment. The experimental equipment has been described in previous progress reports. Following is a list of references in which detailed descriptions may be found: (1) TOP Chamber - 1000 cubic foot chamber for large scale tests. (a) Specifications, pages 25, 27, Progress Report No. 5. (b) Layout drawings, page 28, Progress Report No. 5. (c) Photograph of TOP Chamber, page 35, Progress Report No. 6. (2) Description of light source, photoelectric cell, amplifier circuit, and Esterline Angus Recorder, pages 39, 40, Progress Report No. 10, Quarterly Report No. 7. (3) Piping diagrams for the FS-H20 system, and high pressure spraying system, page 42, Progress Report No. 10, Quarterly Report No. 7. (4) Nozzles: (a) Used with water and kerosene, page 29, Progress Report No. 5. (b) FS-H20 nozzles, page 33, Progress Report No. 6. II Experimental Procedure. The nozzle to be used for a particular test was fitted to the outlet located at the middle of the chamber's south wall and near the floor so as to spray in a direction of 45~ to 60~ with respect to the floor. Any smoke remaining in the chamber from a previous test was blown out with a compressed air jet. The zero and initial light intensity readings of the Esterline Angus Recorder were nted, and the corresponding values in foot candles were obtained from a Photovolt photometer. The fluid to be sprayed was charged into the positive displacement chamber from the reservoir by compressed air. The volume of fluid in the displacement chamber was measured with an indicator gage calibrated in 50 ml divisions Pressure was applied to the positive displacement chamber piston by turning on the Simplex pump. With the valve to the spray nozzle outlet in the closed position, the desired spraying pressure was obtained by manually adjusting the pressurecontrolling, by-pass valve. To initiate spraying the valve to the spray nozzle

29 UNIVERSITY OF MICHIGAN Engineering Research Institute outlet was then opened. Any fluctuations in line pressure were then quickly adjusted to the desired value (time required, approximately two seconds). The spraying time was measured by means of a stop watch. The smoke in the chamber was then allowed to settle, and a "decay curve" of the light transmitted vs. time was recorded automatically on the Esterline Angus recorder. The FS-H20 mixtures were sprayed from special nozzles. These mixtures were sprayed under a pressure of 60 psig supplied from a compressed air outlet. The flow rates were regulated by manually adjusting the discharge valves on both the FS and water linsa. The milliampere readings of the Esterline Angus Recorder were converted to foot candles by means of calibration curves. These curves were obtained by spraying 600 millileters of kerosene into the chamber through nozzle F-80NS-0.60-80~M at a pressure of 1000 psig and by making simultaneous readings of the light transmitted through the settling smoke on a Photovolt photometer and the Esterline Angus recorder. Figure 14 gives typical calibration curves for the recorder. Due to peculiarities in the amplifier, calibration curves were necessitated at frequent intervals. In an attempt to minimize variations in the behavior of the circuit, the light source, the amplifier, and the recorder are now operated from a Sola, 115-volt, #3086, constant voltage transformer. Further improvement in this direction could be made by redesigning the amplifier to give a linear response to the input signal. TYPICAL CALIBRATON 0.9 CURVE FOR ESTERLINE ANGUS RECORDER. ioc 0.U i i U (4.0 -! 0 \\ PHOTOMETER READO ING FOT-CNDLES Figure 14.

30 UNIVERSITY OF MICHIGAN Engineering Research Institute III Experimental Data and Summary of Results. A total of 60 decay curves for kerosene, water, and FS-H20 smokes were made in order to evaluate the smokes produced by atomization and smokes produced by the FS-H20 method. The results presented herein are based on smokes produced by Monarch Company nozzle type F-80NS-0.60-80PM, which was used to atomize kerosene and water at a spray pressure of 1000 psig. The FS-H20 mixtures were sprayed from special nozzles by compressed air at a pressure of approximately 60 psig. Figure 15 is a plot showing the per cent light transmitted through the smokes versus the decay time in minutes. This plot was made from the original curves obtained on the Esterline Angus recorder by converting the milliampere readings to foot candles of light intensity with appropriate calibration curves. The volume of fluid sprayed, the parameter for each curve, is shown on the plot. The upper curve, labeled FS-H20, is taken from the decay curve of the FS-H20 smoke and is used as a "standard" (to be explained later). Figure 16 is a plot correlating the minimum value of the fractional light transmitted through the smokes versus the volume of fluid sprayed into the chamber. Table I is a tabulation of data obtained from the decay runs and used to compare the smokes. Table II gives the calculated average effective particle diameter as a function of the volume of fluid sprayed. These calculations will be discussed in the following section. Table III shows the weights of kerosene and water necessary to produce a smoke equivalent to the FS-H20 "standard" smoke.

31 UNIVERSITY OF MICHIGAN Engineering Research Institute 10 FS,- H0 2 0 STANDARD 20 830 ^ A 40 - \ __ _ _ _ RP \MR ON OF LG T CRIFES co \ \ ~^ 1 80 500 1H 1`1 ALF LFE 4 I 40 0 Il RY VR SSU.M. K E R SSENE X 650 ml H20 Li NOZZLE F-80-NS-0.60-80~ M SPRAY PRESSURE 1000 PSI oT 500 ml H20 g O r20 30 40 50 60 70 DECAY TIME-MINUTES COMPARISON OF LIGHT TRANSMISSION CURVES FOR VARIOUS SUBSTANCES Eigure 15.

32 UNIVERSITY OF LICHIGAN Engineering Research Institute MINIMUM FRACTION OF LIGHT TRANSMITTED BY WATER AND KEROSENE AS A FUNCTION OF VOLUME SPRAYED. 0.9 NOZZLE F-80NS-0.60-80~M t- 0.8 \" _. SPRAY PRESSURE 1000 PSIG 0 100 200 300 400 500 600 700 E SPRAYD I MILLIT-K EROSENE i0.6 ge 16. The curves of Figure 15 permit a qualitative comparison to be made between the smokes produced by Monarch Company nozzle F-80NS-0.60-800MI and those by the FS-H20 method. The curves shown have the general characteristic of approaching the abscissa asymptotically as time increases. It is seen that the FS-II20 standard smoke produced by a total charge of 28 ml of fluid gives a more intense and stable smoke than either water or kerosene. The half-life time of decay for these smokes (Progress Report Nfo. 6, page 22) gives a qualitative measure of their stability. For the FS-H20 standard the half-life time is approximately 250 minutes as compared with 50 minutes for a 500 ml charge of kerosene and 7.5 minutes for 500 ml of water. (A correlation of half-life time is not included in this report since the data at present are ~~~0._ _|incomplete.i) 0 437

TADIZA I. Experimental Data Io Ip Minimum Volume Spray Initial Minimum % Light Run Material Sprayed Pressure Intensity in Value in Transmission Number Nozzle Sprayed ml_ psig foot candles. Foot Candles I./In x 100 F-31 F-80NS-0.60-80~M Water 650 1000 44.2 12.48 28.2 F-32''' " 650 4 44.2 13.22 29.9 F-8 9 n 4 6 500 n 46.0 17.4 37.8 F-9 " n " n 500 4 45.5 15.6 34.3 F-33 "'" 400 44.8 18.2 40.6 F-43 "'9 " 400' 43.0 18.3 42.6 F-44 n'' " 400 40 43.0 18.3 42.6 F-34 3' 300 * 44.7 21.05 47.0 F-35 0 n " 300 33 2 47.0 F-56 n'' 2030 1 43.0 26.0 61.0 F-57 n " n 200 " 43.0 26.7 62.1 F-50''' U of Mich 500 9 43.0 5.93 13.8 kerosene F-51 *'9' 500' 43.0 5.93 13.8 F-52''' " 500 43.0 6.0 13.95 F-46 *' " * 400 w 43.0 7.82 18.2 F-47 2' 400 * 43.0 6.5 15.1 F-48'''* 400 * 43.0 7.82 18.2 F-49'' w' 400' 43.0 8.3 19.3 F-53'''' 300 ^ 43.0 12.5 29.1 F-54'" " " 9 30 430 0 11.9 27.7 F-55'' n' 300' 43.0 11.6 27.0 F-22'' 200' 47.3 18.02 38.1 F-26 " " " 200 " 48.0 18.0 37.5 F-23' n''' 100' 51.5 30.0 58.3 F-27 " 100 49.5 29.0 58.7 F-60 FS - water nozzle FS-H20 28 60 39.5 13.5.8 15J "5.3

34 TABLL II Comparison of Effective Average Particle Sizes for Water and for Kerosene Volume of Fluid Calculated d Calculated d Sprayed into in BMicrons for in M},icrons for Chamber in ml. Kerosene by Eq.(24) dater by Eq_ (24) 650 -- 55.1 500 30.5 51.4 400 28.0 48.3 300 28.4 41.7 200 25.0 43.2 Average 28.0 Average 48.0 TABLE III Relative -Jeights of Kerosene and W-ater Required to Produce an Equivalent Smoke Based on the Standard FS-H20 Smoke. Calculated d Total Volume Weight of Weight Ratio Equation (24) of Material Mtaterial wt. of materit Material in in Cubic ir wt. of FS-HO2 Sprayed Microns Centineters Gr as standard FS-water standard 13 + 15 m1. 1.5 28 40 1 U. of Michigan kerosene 28 755 617 15.3,dater 48 1493 1493 37.4 Assuming a possible d = 10 microns from the vibrating type nozzle. U. of CMichigan kerosene 10 270 221 5.5 ~Vater 10 311 311 7.8

35 UNIVERSITY OF MICHIGAN Engineering Research Institute IV Discussion of Results. The equations of the curves in Figure 16 correlating the minimum fraction of transmitted light with volume of fluid sprayed into the chamber are of the form I = e - kS Io (18) where I ) = minimum value of the transmitted light intensity at the end of p spray in foot candles Io= initial incident light intensity in foot candles S volume sprayed into the chamber in milliliters k = a constant for a particular nozzle anld fluid under constant spraying conditions Equation (18) is of the same foir as the equation that relates the fraction of the light transmitted to the particle cross-sectional area, the concentration, and the length of the path. This equation which neglects forward scattering is written: Io I ~- = e" - (1) where I = transmitted light intensity in foot candles Io = incident light intensity in foot candles q = scattering cross-sectional area of the particle in square centimeters n = number of particles per cubic centimeter 1 = length of the path in centimeters For purposes of comparison, let q equal the cross-sectional area of an average "effective" particle 7rd2 q = (19) where d = average "effective" particle diameter in centimeters Another equation nay be written to relate S, the volume of fluid sprayed into the chrmber, to n, the number of particles per cubic centimeters having a diameter d, as defined above.

36 UNIVERSITY OF MICHIGAN Engineering Research Institute S = vn C where S 3 volume of fluid sprayed into the chamber in cubic centimeters v s volume of particle having diameter d, -= d in cubic centimeters 6 n = number of particles per cubic centimeters c = volume of the smoke chamber Substituting for v in equation (20) and solving for n gives S 6S vC Vrd3C (21) Substituting for n and q in equation (1) and solving for d gives, 313S d = 2 C In Io (22) Ip 3 1 The quantity 2 C is a constant determined by the dimensions of the spray chamber. For the chamber 1 is 10 feet and C is 1000 cubic feet. Let k =31 = 3(10)(12)(2.54)(104) microns _- 0.1615 2 c 2(10)5 (12)3 (2.54)3 cm Substituting for k in equation (22) 0.1615 S d =.. In I (23) Io Equation (23) permits an order of magnitude calculation of an average.ffective particle diameter from the volume of fluid sprayed into the chamber and the experimentally determined values of I and Io. If the measured volume of fluid sprayed into the chamber is substituted into equation (23), the value of d calculated will not be a representative average of the conditions in the chamber at the end of the spray period. An analysis of the spraying process shows that a certain fraction of the fluid sprayed is not in the dispersed form at the time the minimum value of the transmitted light is read. Due to incomplete atomization, some fluid rapidly settles in the immediate vicinit;

37 UNIVERSITY OF MICHIGAN Engineering Research Institute of the nozzle. Of the portion that is atomized, a certain amount of the coarser particles are continuously settling out of the system. The magnitude of this second loss is a function of spraying time and the size and number of particles produced by the nozzle for a given fluid at constant spraying conditions. Some tests were conducted to determine the fraction of fluid lost due to incomplete atomization. The results of these tests are tabulated in Table IV. These data were obtained by absorbing the fluid that settled out on paper over an area of four feet square under the nozzle and determining the weight increase immediately after spraying. The average per cent loss for kerosene was 25 per cent and for water, 35 per cent, or an effective fraction of fluid sprayed equal to 0.75 and 0.65 for kerosene and water respectively. Thus, if "f" equals the effective fraction of fluid sprayed and applying this correction, equation (23) becomes, 0.1615 f S d = In I (24) Ip where d = average effective particle diameter in microns. Table II is a tabulation of diameters calculated from equation (8) using the experimental data listed in Table I. The variations of calculated diameters is 21.8% for kerosene and 24.3% for water based on the smallest calculated diameter. Theoretically, for a uniform dispersion of particles in the light beam, if the calculated average diameters were plotted against the quantity (f x S) and the curve were extrapolated to (f x S) equal to zero, the diameter so obtained should be more representative of the effective average diameter of the particles which leave the nozzle in the atomized state. This extrapolation affords a means for evaluating the properties of the smoke at zero settling time and should tend to eliminate the effects of coalescence and rapid settling rates of the larger particles. Table V is a compilation of data obtained with the present experimental equipment and technique. An examination of the d's calculated for kerosene, runs P-21 through P-31 reveals comparatively good reproducibility and seems to indicate that the calculated diameter is independent of the volume sprayed. This behavior may be possibly attributed to the fact that the particles produced in atomizing kerosene fall within a very narrow size distribution (uniform particles) of such a small diameter that they do not settle appreciably during the spraying times involved in these tests. Obviously, this assumption is subject to the limits of accuracy of the experimental methods used.

38 UNIVERSITY OF "MICHIGAN Engineering Research Institute TABLE IV Fractions of Fluids Sprayed hich Settle Out Instantaneously Volume we ight Equivalent Volume of water of water volume of % Effective sprayed lost in lost water water fraction- Average in ml. _ grams. ml. lost. f. f. 600 234 234 39.0 0.610 600 206 206 34.3 0.657 500 178 178 35.6 0.644 500 149 149 29.8 0.702 400 142 142 35.5 0.645 300 99.4 99.4 33.1 0.669 200 71 71 35.5 0.645 0.65 Volume Jleight of Equivalent Volume of kerosene kerosene volume of % Effective sprayed - lost - lost kerosene kerosene fractionml. _gr as. ml. lo st. f. 600 135 167 2708 0.722 500 121 149.5 29.9 0.701 500 92.2 114 22o8 0.772 400 78 96.5 24.1 0.759 300 63.9 79 26.3 0.737 200 35.5 43.8 21.9 0.781 0.75 Area of paper = 4 feet square

I J.ADL4-a V Experimental Data on Nozzle F-80MS-0.6 0-80~M at a raying Pressure of 1000 Psige Calculated Volume Spraying 10 in Ip in diameter Run sprayed time in Material foot foot pin microns Number in ml. seconds, sprayed, candles. candles. I0 Equation (24) P-i 500 375W ater 42.8 15.9 0.372 5.1 P-2 500 385 43.1 16.2 0.375 5.5 P-3 400 305.2 42.9 17.3 0.403 46.2 P-4 400 305.3 42.8 16.9 0.395 45. P-5 300 225.5 42.8 22.0 0.514 47. P-6 300 227.8 42.9 21.5 0.502 45.7 P-7 200 151.8 42,9 24.8 0.578 38. P-8 200 152 42.9 24.0 0.560 36.2 P-9 100 68.6 43.1 35.6 0.826 54.8 P-10 100 69.8 42.9 33.8 0.788 44.3 P-11 100 68.8 42.9 35.5 0.827 55.1 P-12 100 69.4 42.8 37.8 0.883 84.6 P-13 100 62.9 46.2 41.0 0.88787.3 P-14 150 110.7 46.5 35.0 0.752 55.2 P-15 150 110.9 46.0 35.0 0.752 55.2 P-16 150 110.8 46.8 26.2 0.774 61.4 P-17 150 112.4 46.8 33.3 0.712 46.1 P-18 150 112.2 46.5 35.0 0.752 55.2 P-19 200 1b2.3 45.8 32.5 0.710 61.2 P-20 200 154 45.6 32.8 0.718 6.4 P-21 100 45.4 kerosene 42.8 31 0.725 50.2 P-22 100 51.3 43.1 26.5 0.625 257 P-23 100 53.5 43.6 29 0.664 29.6 P-24 100 53.5 43.1 26.5 0.625 25.7 P-25 100 53.5 42.2 26.0 0.616 25.1 P-26 100 55.8 41.8 27,5 0.658 29.0 P-27 300 183.7 42.0 10.2 0.243 25.7 P-28 300 184.7 43.6 10.8 0.247 25.8 P-29 400 247.8 43.0 7.1 0.166 26.9 P-30 400 249.4 42.0 6.8 0.612 26.7 P-31 500 310.4 43.2 4.8 0.111 27.7 P-32 200 150.l Aater 45 29.1 0.648 49.1 (continued on next page)3 I o

o TABLE V (continued) Calculated Volume Spraying 10 in Ip in diameter Run sprayed time in Material foot foot in microns Number in ml. seconds. sprayed. candles. candles. Io Equation (24) P-33 200 150.8 Water 44.9 28.5 0.635 4.63 P-34 200 1514 44.8 31 0.692 57.0 IP-35 300 230.5 44.8 24.5 0.547 52.2 P-36 300 233.5 45.0 22.5 0.500 45.5 P-37 300 235.6 44.8 22.3 0.497 44.2 P-38 400 305.6 44.2 18.8 0.425 49.1 P-39 400 307.6 44.2 20.0 0.452 52.6

41 UNIVERSITY OF MICHIGAN Engineering Research Institute In general, the calculated diameters for the water runs, P-1 through P-20 and P-32 through P-39, show a greater degree of scattering than the kerosene runs. The degree of scattering seems to be more pronounced for the lower volume of water sprayed. Visual observation of the smokes indicates that the water smokes contain a greater proportion of coarse particles than kerosene smokes. (This observation is confirmed by the calculated values of "d" for water, which are all larger than the corresponding values for kerosene.) These results seem to indicate that there is a non-uniform dispersion of particles in the light beam at the end of the spraying period, when the transmitted light intensity is measured. Considering the particles in the chamber from a statiStical viewpoint, the probability of reproducing a given distribution from a fluid having a wide-peaked distribution curve is lower than the probability of reproducing a given distribution from a fluid which has a narrow-peaked distribution curve Preliminary tests with the FS-H20O mixtures showed that it would be necessary to spray very small quantities of fluid to obtain a readable transmitted light intensity on the photometer and Esterline Angus recorder. A mixture of 13 ml of FS and 15 ml of water when sprayed into chamber gave a ratio of Ij equal to 0.038. This mixture was arbitrarily selected as a Io standard to make quantitative comparisons of water and kerosene smokes. SubstitutioL of these experimental values in equation (23) gives a calculated d"w of 1.5 microns for FS-H20. Substituting the standard value of Idp 0.038 and the average 1o effective diameters from Table II in equation (24), the value of S obtained represents the amounts of water and kerosene necessary to give the same per cent light transmission as the FS-water standard. The results of these calculations for kerosene and water are shown in Table III. Assuming an average d of 10 microns from the vibrating type nozzles gives a weight ratio of 5.5 and 7.8 for kerosene and water respectively. It is believed that with further improvement of the nozzle design, a weight ratio between 3 and 4 can be realized. V Conclusions. As was expected, the FS-H20 mixture for producing smoke is superior to the direct atomization of fluids on a pound for pound basis. However, it is believed that this ratio can be improved corsi.erably. For example, as pointed out in Figure 22, Progress Report No. 6, a S, 3bler-Kempf kerosene is approximately twice as stable as a U. of Mich. kerosene. It is conceivable that other fluids would give still better results. The methods employed in this section for evaluating the effective average particle sizes in a smoke are only approximate. However, it is felt that the results give to a degree a quantitative basis for comparing smokes. The absolute reliability of these results obtained in the TOP chamber by means of light transmission data, only, cannot be established until comparisons are made with corresponding data obtained from the refined measurements on both scattered and transmitted light, which were discussed in Part A of this report.

42 UNIVERSITY OF MICHIGAN Engineering Research Institute Part C Vibrating-Type Atomizing NozzleA As reported in previous progress reports (Progress Report No. 5, page 28 and Progress Report No. 10, Quarterly Report No. 7, page 23) preliminary investigation of the vibrating-type atomizing nozzle indicated that smaller and more uniform particle sizes might be obtained with this type of nozzle than with any commercial nozzles. It was also observed that the operating characteristics of the vibratingtype nozzle are rather sensitive to changes in the design variables, among which are (1) Frequency of vibration of the spring (2) Initial tension in the spring as related to the spraying pressure (3) Diameter of the orifice (4) Clearance between the valve stem and orifice (5) Included angle of the tapered portion of the head of the stem which seats on the orifice (6) Ratio of the diameter of the head of the stem to the orifice diameter (7) Surface finish of the component parts of the nozzle. Subsequent tests have indicated that the relative "stiffness" of the spring or the spring constant is also a factor. For the purposes of an initial testing program, it was decided to study the relationship between capacity and spraying pressure for parameters of spring tension, spring constant, and the clearance between the valve stem and orifice. Throughout these tests, qualitative observations of the other variables were noted. I Experimental Equipment. All tests were conducted in the TOP building, using the spray equipment previously described in Progress Report #6, pages 34 ff, and Progress Report No. 10, Quarterly Report No. 7, pages 39 ff. In order to measure the initial tension on the spring, a mechanical loading device was designed and built, (Figure 17). To determine the spring tension, the nozzle is first threaded into its seat, and the dial gage is positioned so that its stem contacts the head of the nozzle stem. An ordinary postal scale is placed under the nozzle. The bracket supporting both the nozzle and the dial gage may then be lowered by turning the knurled head on the threaded rod until the bottom end of the stem contacts the scale. As the bracket is lowered further, the scale pan is depressed and the valve stem is moved upward. The deflection is indicated on the dial gage. Typical load-deflection curves are shown in Figure 18. As can be seen from this figure, the point at which the curves break sharply is a function of the initial tension on the spring and is defined as the spring tension.

43 UNIVERSITY OF MICHIGAN Engineering Research Institute MECHANICAL LOADER FOR DETERMINING INITIAL SPRING TENSION Figure 17. TYPICAL LOAD DEFLECTION CURVES FOR SPRING NO. 2 (SPRING CONSTANT IS DEFINED AS SLOPE OF LINE) 110 100 90 - 80 0 -0' 70:3 60 s''S —'o,i _ -, __ l 0 ______, _'________ _o. -J 00.O0.002.004.006.008.010.012 DEFLECTION (INCHES) Figure 18.

44 UNIVERSITY OF MICHIGAN Engineering Research Institute The slope of the flatter portion of the curve is a measure of the *stiffness" of the spring and is defined as the spring constant. The variation of the spring constant for Spring No. 2 with initial tension, as shown by Figure 18, is representative of the experimental accuracy or reproducibility which is obtainable from the Mechanical Loader of Figure 17. The type of nozzle originally planned for these tests was equipped with a leaf spring set into a recess in the bottom of the nozzle body (Figure 23, Progress Report No. 5). Initial tests with this arrangement indicated that the leaf spring so fixed the position of the bottom of the stem that it was difficult to center the stem in the orifice during spraying. By replacing the leaf springs with a helical spring surrounding the valve stem, as shown in Figure 19, much better centering of the stem in the orifice was obtainable. In this type the valve stem is suspended freely; therefore, it will tend to *float" in the fluid stream and to seek its own center. The data presented in this section were obtained from the helical spring model, rather than the leaf spring. Since the clearance between the orifice and the stem was one of the variables to be studied, four stems of different diameters were made up for Nozzle No. 17, which has an orifice diameter of 0.040 inches. The stem diameters were: 0.0277, 0.0300, 0.0320, and 0.0358 inches. After considerable usage it was noted that the 0.0358 inch stem had become scored. The stem was refinished and its new diameter is 0.0338 inches. Another 0.0358 inch stem was also made. aoor,, A ctHi ss eua VC.., < - o.^^C-. 5,ZE Fi gure 19.

45 UNIVERSITY CF MICHIGAN Engineering Research Institute The ratio of the diameter of the head of the stem to the diameter of the orifice and the included angle of the head of the stem (1000) which seats on the orifice were held constant for all stem sizes. For comparative purposes, several runs were also made while using Nozzle No. 13 (orifice, 0.0700 inches in diameter) and stem No. 9 (0.0630 inches in diameter). Time did not permit a more exhaustive study of this nozzle-stem combination. II Experimental Procedure. After assembling the nozzle, stem, and spring, and determining the spring tension, the nozzle was placed in position in the spray chamber. A ranging" run was first made; i.e., the line pressure was gradually increased, and a record was made of the pressure at which spray started, the pressure at which audible Vibration started, and the maximum pressure obtained. No attempt was made to quantitatively define the point at vwich spray started. In general the cone built up gradually, and spray was not considered to have started until the cone was fairly full. Although the definition of this point varied with various observers, the data taken on minimum spraying pressure correlates reasonably well (see Figure 20). After two or three runs, usually at higher pressures, the nozzle was removed from the spray manifold, and the spring tension checked. (It had been found that the vibration sometimes loosened the knurled nuts and thus changed the spring tension.) Runs were made at pressure increments of 50 lbs/in2, two runs being made at each pressure. 500 ml of water was sprayed for each run, and the time of spray was measured with a stop watch. Also recorded were any observations on the cone and the spray characteristics, the relative pitch and intensity of vibration, and any other factors which might possibly affect the validity of the data. At the end of the series of runs, the spring tension was again checked. Initial runs indicated that centering of the stem is an important factor in producing audible vibration. Often a run was made without vibration, to be followed by a vibrating run at the same pressure. Investigation showed that in such cases,vibration could be induced by forcibly holding the stem down against the orifice until a slight back pressure was built up. When the stem was released, the impact of this pressure centered the stem, and vibration occurred. It appears that a vibrating stem remains centered as only rarely did audible vibration stop once it had started, except at lower "critical" pressures. It was subsequently discovered that vibration could also be induced by tapping the side of the nozzle body after the spray had started, the force of the blow evidently being sufficient to momentarily center the stem and permit vibration to occur. Since "off-centered-stemw runs generally produced half, or two-thirds spray-cones, even at those pressures at which vibration did not occur the tapping procedure was also followed on non-vibrating runs, so as to keep the stem centered and provide comparable non-vibrating data.

46 UNIVERSITY OF MICHIGAN Engineering Research Institute III Experimental Data and Results. The theoretical relationship between the initial spring tension and the minimum pressure required to produce a spray may be easily derived in the followin6 manner: Tension in the spring = (minimum spraying pressure) (Area of Orifice) T = (P) (0.785) (d2) (16) T = 12.58 Pd2 (25) where T = initial spring tension in ounces P = minimum spraying pressure in psi d - diameter of orifice in inches For nozzle No, 17, with an orifice diameter of 0,04 inches, Equation (25) becomes: T = 12.58 (0.04)2 (P) = 0.201 P (26) For nozzle No. 13, with an orifice diameter of 0.07 inches, Equation (25) becomes: T = 12.58 (0.07)2 (P) = 0.616 P (27) On Figure 20 the deviation of the experimental data from the theoretically established relationship is shown. The scattering of the data is probably due to several factors: (1) Failure to sharply define the minimum spraying pressure (see page 45 ). (2) Errors in the original measurement of the diameter of the orifice and from any subsequent dimensional changes resulting from continued usage. (3) Errors in determining the initial spring tension (see page 42 ). The theoretical relationship between the capacity of a nozzle and the spraying pressure is approximately given by the familiar orifice equation: Q= (30) (i ) (do + d5) (do - d+) (CO) (28)

47 UNIVERSITY OF MICHIGAN Engineering Research Institute 35 / THE RELATIONSHIP BETWEEN SPRING / TENSION AND MINIMUM SPRAYING OA; PRESSURE. 1 30- v -/ — v — /0 o ~ 220. W A deI aA E ~ ~~/ *. < /8 +, do =i t of/ /ic 1he5^9 caact VIBRATING NOZZLE N17 / * I / * STEM DIA. IN SPRING CONSTANT INCHES IN LB/INCH I 40.0358 13.7 |o - - ---- e * ----— 10 —------- 0.0358 47.9 /. 0358 76.7 /A A.0358 / isblee A.0320 13.7 Ah c i 0di.0320 76.7 5.0300 13.7 /is +.0277 13.7 orfc seto isntsap ltga n or* NOZZLE 3, SPRING o + DATA FROM TABLE EI P = spraying pressure in psi do = diameter of orifice in inches d s diameter of stem in inches C = coefficient of discharge The capacity versus pressure curves for Nozzle No. 17 (orifice = 0,04 inches) and a stem 0.052 inches in diameter, with initial spring tension as a parameter, are represented in Figure 21, The point of intersection of these curves with the zero capacity axis are taken from Figure 20, For pressures above 1500 psi, the curves for the three initial spring tensions become identical. It is believed that the 1500 psi pressure represents the point at which the head of the valve stem has assumed a position in the fluid stream sufficiently removed from the orifice seat that it can no longer affect the flow characteristics through the annular discharge. The theoretical curve is a plot of equation (28), assuming that the coefficient of discharge, C, is equal to one. On this basis, the coefficient of discharge for nozzle No. 17 with a stem 0.032 inches in diameter is approximately 0.91. This value is entirely plausible since the entrance to the orifice section is not sharp, but gradual, and more nearly approximates the behavior of a rounded-edge orifice. (See Figure 19.)

0.80 CAPACITY vs PRESSURE FOR NOZZLE NO. 17 0.032 INCH STEM. THEORELTICAL 0URVE~, J 060 i0.50 --- 0a Ie ( 120I —^- 4 Z. OZ. 1.5 OZ_.. _._ SPRING CONSTANT- 13.7 LBS/IN. J / I U-MINIMUM SPRAY PRESSURE FIG 20 g010 I - OZ.- SPRING TENSION I I' DATA TABLE —' 0 2 00 K400 12SW 1400 IBO 1800 SPRAYING PRESSURE PSIG516 Figure 21.

49 UNIVERSITY OF MICHIGAN Engineering Research Institute Similar curves are represented in Figures 22 and 23 for valve stem diameters of 0.0338 and 0.0358 inches, respectively. Although these curves do not become identical above a certain pressure as was observed at 1500 psi in Figure 21, it is believed that the average deviation (roughly, 3 per cent) lies within the experimental accuracy of the methods employed. It is apparent from the 18 oz. curve of Figure 22 that the behavior of the nozzle under the conditions of operation was rather erratic. The sharp break in this curve at 1200 psi down to the minimum spraying pressure may have been caused by some obstruction in the annular discharge or failure of the stem to center properly. The fact that the actual capacity exceeds the theoretical capacity (discharge coefficient equal to one) for the curves of Figure 22 is not uncommon Similar phenomena have been observed in the case of diesel fuel injection valves and may be attributed to'surging". In general, at higher pressures where the capacity versus pressure relationship bedomes independent of the initial spring tension, the coefficient of discharge lies between 0.90 and 0.93. Attempts to correlate the flow characteristics of the nozzle with the frequency of vibration of the spring have not been successful to date. Figure 24 is a replot of Figures 21, 22, and 23, and compares the flow characteristics of the nozzle for various stem diameters and initial tensions in a spring having a spring constant equal to 13.7 pounds per inch. For a given spring constant it is reasonable to expect that the curves of capacity versus pressure for a fixed initial spring tension will coincide for all stem diameters in a given orifice up to the pressures where the slope of these curves begins to decrease rapidly. Figure 25 illustrates the effect of spring constant on the capacity versus pressure relationship while holding all other variables (stem diameter, orifice diameter, and initial tension) constant. For a given spring it appears that the rate of change of capacity with pressure, at the lower pressures, is independent of the initial spring tension. (See Figure 24 also,) The data from itich the curves of Figures 20 through 25 were obtained are presented in Tables VI through XI.

0.8 CAPACITY vs. PRESSURE FOR NOZZLE NO. 17, 0.0338 INCH STEM l.I THEORETICAL OURVE ____0._7 |. SPRING CONSTANT = 13.7 LBS./INCH ____ ___ 0.7 I-MINIMUM SPRAY PRESSURE FIG. 20 OZ. -- SPRING TENSION DATA TABLE vrrr 15 0.6 H d i O. 0.4 0 03 i) 0.2 / S / / A/ / O10 OZ. 1/ O 5OZ. /180Z. / 26 OZ. q / f/ 0.1, / 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 2800 3000 SPRAYING PRESSURE PSIG Figure 22, 0 to

'0.7 1 F i CAPACITY vs. PRESSURE FOR NOZZLE NO. 17, 0.0358 INCH STEM 0.6 SPRING CONSTANT =13.7 LBS/INCH *-MINIMUM SPRAY PRESSURE FIG.20 OZ- SPRING TENSION 0.5___ L-_ ______ ______ _____DATA TABLE r X_ __________ g THEORETICAL CURVE P o0~~~~~~~~.1~ ~ ~ ~ ~ ~ / /, / I 0 I / I I11 I 0 _ _ _ _ _/9 OZ. 214 OZ. / 18 OZ. *25.5 OZ. 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 SPRAYING PRESSURE PSIG Figure 23. $-J

COMPARISON OF CAPACITY vs. PRESSURE ETICAL CURVE FOR.032 STEM FO NOZZLE NO. 17, ORIFICE 0.040 IN. TH I C FOR VARIOUS STEM DIAMETERS ~ THEORETICAL CURVE FOR 0.0338" STEM 0.7 ~~~~~0.7 t lSPRING CONSTANT 13.7 LBS/ INCH OZ.-SPRING TENSION -_ REPLOT OF FIGURES 21,22,23 — ( D0.6 W z 2 0.5 --------— H — Z. ---- ^ - THEORETICAL CURVE FOR 0.0358" STEM 9 OZ. 18Z. 25.5 OZ.O oOZ. _ _ _I__ ________ _ _ SPRAYING PRESSURE PSIG Figure 24. tC 100. 0

a40( ~ -I 1 COMPARISON OF CAPACITY VS. PRESSURE FOR NOZZLE N0.17 STEM 0.0358 IN. FOR VARIOUS SPRINGS K=SPRING' CONSTANT IN LBS./IN. 0.3 _ _______ ~___~~___DATA-TABLEX AND TABLE 3L Q33 Q,30 Km 47.9 33 OZ. TENSION Lzj 0.25 K-2355 33.5 OZ. TENSION i I I -J_ _ _ I _ _d Q20 K-H —5KOZ -T IS I O N 14.5 OZ.TENSION I K 13.7 |,' 5. IQN 4~~~~~ -3 IH 1 Z T N IN/ 5.5 OZ. TENSIONOTE N / 0.15 ---- - -- --- -0-T ---------------' ------------ I'/Cl // s~~~~~~~~~~~~~~ 0.0 5 5.5 OZ. TENSION~1, / I I II /I I / _____ _ _______1I I_________ I I____________ 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 2600 SPRAYING PRESSURE - PSIG Figure 25.

54 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VI Effect of Initial Sprig Tension on Minimum Spraing Pressure; Nozzle No. 17 (Orifice = 0.04 inches), Various Stems and Springs, Stem Spring Spring Minimum Run Diameter Constant Tension Pressure No. Inches lbs/in. Ounces Ibs/ in2 617.0358 13.7 5 175 1254 " 200 1181 " 300 561 6 250 163 8.5 375 716 10 400 674' 450 1108' 550 1371 10.5 400 1372 " 500 1332 " 600 719 11 575 824 13 550 136 13.5 600 400' 750 893 15 850 966 17 775 135 18.5 900 65 32 1660 295.0358 47.9 15 650 181 33 1500 251.0358 76.7 5.5 200 222 9.5 400 207 14.5 725 281.0358 14.5 625 493.0320 13.7 4 125 997 2 200 "1T6 4.5 200 544 5 175 471 8 300 447 " 400 510 8.5 350 339 8.5 400 1012 9 400 311 14.5 700 1049 15 700 319 19 825 47 27 1350 34 32 1550

55 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VI (continued) Stem Spring Spring Minimum Run Diameter Constant Tension Pressure No. Inches lbs/in. Ounces lbs/in2 345 33 1525 353.0320 76.7 18.5 775 351 31 1450 1203.0300 13.7 5 300 29 12 550 1201 16 800 25 19 875 17 27.5 1350 13 31.5 1620 1251.0277 13.7 1.5 75 1243 5.5 250 5 16 750 4 19 950 12 27 1300 Comparison runs with Nozzle No. 13, Stem No. 9 (Orifice = 0.07 inches). 357 76.7 15 300 197 29 850

56 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VII Effect of Spring Tension and Spraying Pressure on Capacity; Nozzle No. 17, 0.032 inch Stem. Spring Spring Spraying Spraying Run Constant Tension Pressure Time Capacity No. lbs/in. Ounces lbs/in2 Seconds Gal/min. 502 13 7 4 250 37.3 0.212 501 300 33.0 0.240 498 350 28.1 0.282 496 400 24.2 0.327 494 450 21.3 0.372 1009 800 16.5 0.480 1008 850 16.0 0.495 998 925 15.4 0.514 999 950 15.1 0.524 1005 980 14.9 0.531 010 1460 12.7 0.625 521 13.7 8 500 40.6 0.195 518 575 29.0 0.273 1046 600 26.8 0.295 1045 650 23.0 0.344 1042 700 20.4 0.388 1039 750 18.1 0.438 1037 800 17.7 0.448 1035 850 16.9 0.469 1033 900 16.3 0.486 1031 950 15.7 0.505 1029 1000 15.3 0.518 1028 1050 14.9 0.531 1026 1100 14.5 0.546 1023 1150 14.2 0.558 1021 1200 13.9 0.570 1020 1250 13.7 0.579 1017 1300 13.4 0.591 1015 1350 13.2 0.600 1094 13.7 14.5 850 36.4 0.218 1092 900 28.4 0.279 1090 950 24.0 0.330 1089 1000 19.7 0.402 1087 1050 17.6 0.450 1085 1100 16.0 0.495 1082 1150 15.0 0.528 1080 1200 14.5 0.547 1078 1250 14.0 0.566 1076 1300 13.6 0.583 1074 1S50 13.3 0.596

57 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VII (continued) Run Spring Spring Spraying Spraying Capacity No. Constant Tension Pressure Time Gal/rain lbs/ in. Ounces Ibs/in2 Seconds__.. 1073 13.7 14.5 1400 13.1 0.605 1070 1450 12.8 0.619 1068 1500 12.5 0.634 1066 1550 12.3 0.645 1064 1600 12.1 0.654 1062 1650 12.0 0.660 1060 1700 11.8 0.671 1058 1775 11.6 0.683 1056 1875 11.3 0.700

58 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VIII Effect of Spring Tension and Spraying Pressure on Capacity; Nozzle No. 17, 00338 inch stem (refinished originally 0.0358 inch stem Spring Spring Spraying Spraying Capacity Run Constant Tension Pressure Time lNo. bs/in. Ounces lbs/in2 Seconds Gal/Min. 1197 13.7 5 1000 17.4 0.455 1198 1100 16.7 0.475 1191 1200 16.1 0.492 1188 1500 14.8 0.535 1184 2550 12.*9 0.615 677 13.7 10 550 47.2 0.168 680 600 34.0 0.233 682 650 26.2 0.302 683 700 21.6 0.367 686 750 19.8 0.400 688 800 19.4 0.408 690 850 18.9 0.419 691 900 18.2 0.435 694 950 17.9 0.442 1161 1050 17.2 0.460 1160 1100 16.8 0.471 1157 1150 16.5 0.480 1156 1200 16 2 0.489 708 1250 15.8 0.501 1152 1300 15.6 0.508 1149 1350 15.5 0.511 1148 1400 15.3 0.518 1146 1450 15.2 0.521 1140 1500 15.1 0.525 1138 1550 14.9 0.532 1136 1600 14.8 0.535 1134 1650 14.6 0.543 1132 1700 14.5 0.546 1130 1750 14.4 0.550 1128 1800 14.2 0.559 1125 1850 14.1 0.562 1123 1900 14.0 0.566 1122 1950 13.8 0.574 1120 2000 13.7 0.579 1118 2050 13.6 0.583 1116 2100 13.5 0.587 1114 2150 13.5 0.587 1111 2200 13.2 0.600 1178 2300 13.0 0.610 1176 2400 12.9 0.615 1174 2500 12.8 0.619

59 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VIII (continued) Spring Spring Spraying Spraying Run Constant Tension Pressure Time Capacity No. lbs/in. Ounces lbs/in2 Seconds Gal in. 1172 13.7 10 2600 12.6 0.629 1170 2700 12.5 0.633 1168 2800 12.4 0.639 11^6 2900 12.3 0.644 1164 2950 12.2 0.650 819 13.7 13 700 65.0 0.122 801 750 30.9 0.256 802 800 27.7 0.286 807 850 23.9 0.332 790 900 21.3 0.372 811 950 19.2 0.412 823 1000 17.8 0.445 785 1050 16.7 0.475 783 1100 16.5 0.480 781 1150 16.2 0.489 825 1200 16.0 0.495 813 1250 15.9 0.498 772 1300 15.7 0.505 770 1350 15.5 0.510 768 1400 15.4 0.514 765 1450 15.3 0.518 763 1500 15.2 0.522 761 1550 15.0 0.528 821 1750 14.3 0.554 965 13.7 15 1000 33.2 0.238 963 1050 27.9 0.284 961 1100 21.8 0.363 959 1150 17.9 0.443 957 1200 16 4 0.483 954 1250 15.9 0.499 953 1300 15.5 0.511 951 1350 15.2 0.521 948 1400 14.8 0.535 947 1450 14.6 0.543 945 1500 14.4 0.550 942 1550 14.4 0.550 941 1600 14.2 0.558 939 1650 14.1 0.562 937 1700 13.9 0.570 934 1750 13.9 0.570 932 1800 13.6 0.583 930 1850 13.5 0,587

60 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE VIII (continued) Spring Spring Spraying Spraying Capacity Run Constant Tension Pressure Time No. lb/in. Ounces lbs/in2 Seconds Gal/Min. 929 13.7 15 1900 13,3 0.596 927 1950 13.2 0.600 920 2000 13.2 0.600 916 2050 13.1 0.605 915 2100 13.0 0.610 911 2150 12.7 0.623 907 2250 12.6 0.629 906 2300 12.6 0.629 904 2350 12.6 0.629 898 2400 12.4 0.639 902 2550 12.3 0.645 924 2875 11.8 0.671 895 2950 11.8 0.671 840 13.7 18 1200 96.4 0.082 838 1300 39,8 0.199 835 1400 26.0 0.305 852 1400 27,0 0.293 833 1500 18.9 0.420 844 1600 16.9 0.532 843 1650 14.2 0.558 831 1675 14.1 0.562 891 13.7 26 1400 108.0 0.073 889 1500 37.2 0.213 886 1550 33.0 0.240 884 1600 25.4 0.312 882 1650 20.1 0.394 880 1700 15.9 0.499 878 1740 14.2 0.557 876 1800 13o3 0.596 875 1850 13.3 0.596 872 1900 13.0 0.610 868 1950 13.0 0.610 866 2000 12.6 0.629 L-5 2050 12.6 0.629 862 2100 12.4 0.640 860 2200 12.1 0.655 859 2275 12.0 0.660 858 2375 11.9 0.666

61 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE IX Effect of Spring Stiffness, Spring Tension, and Spraying Pressure on Capacity; Nozzle No. 17, 0,0358 inch Stem. Spring Spring Spraying Spraying Capacity Run Constant Tension Pressure Time No. lb/in. Ounces lbs/in2 Sec. Gal/Min. 180 13.7 9 750 31.7 0.250 174 800 31.6 0.251 173 900 29.6 0.268 179 1000 28.1 0.282 172 1100 27.5 0.288 171 1300 25.6 0.310 170 1500 23.6 0.336 169 1700 21.3 0.372 168 1900 20.5 0.387 167 2100 19.3 0.411 166 2300 18.2 0.436 165 2500 17.5 0.453 164 2700 16.6 0.478 159 13.7 14 700 68.0 0.117 160 59.1 0.134 158 750 33.0 0.240 155 800 31.1 0.255 152 900 28.4 0.279 150 950 27.9 0 284 147 1000 27.1 0.292 145 1100 25.9 0.306 144 1200 25.0 0.317 142 1400 23.7 0.334 141 1600 22.4 0.354 140 1800 21.2 0.374 139 2000 19.9 0.398 138 2200 18.8 0.422 137 2300 18.0 0.441 120 13o7 18 1050 38.8 0.204 116 1100 29.3 0.270 114 1150 25.9 0.306 112 1200 24.8 0.320 110 1250 24.5 0.324 107 1300 26.3 0.302 104 1400 23.8 0.333 97 1600 23.1 0.343 122 1800 21.6 0.367 124 1900 21.1 0.376 125 2000 20.5 0.387

62 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE IX (continued) Spring Spring Spraying Spraying Run Constant Tension Pressure Time Capacity No. lbs/in. Ounces lbs/in. Sec. Gal/Min. 126 13.7 18 2100 20.1 0.394 129 2200 19.7 0.402 130 2300 19.3 0.411 131 2400 18.8 0.422 133 2500 18.4 0.431 373 13.7 25.5 1300 44.5 0.178 371 1350 27.5 0.288 374 1400 27.3 0.290

63 UNIVERSITY OF MICHIGAN Engineering Research Institute TABLE X Effect of Spring Stiffness on Capacity at Various Spring Tensions and Spraying Pressures; Nozzle No, 17, 0.0358 inch Stem. Spring Spring Spraying Spraying Run Constant Tension Pressure Time Capacity No. lbs/in. Ounces lbs/in2 Sec. Gal/in. 308 47.9 15 800 56.1 0.140 307 900 32.1 0.244 298 1000 27.3 0.287 195 47.9 33 1700 31.2 0.254 194 1800 26.1 0.304 191 1900 20.7 0.383 279 76.7 5.5 300 77.4 0.101 277 400 56.5 0.139 276 500 42.5 0.184 273 600 34.1 0.229 271 700 30.7 0.255 220 76.7 14.5 900 59.4 0.132 219 1000 40.8 0.192 217 1100 34.6 0.226 215 1200 29.0 0.270 282 237.5 14o5 1000 56.7 0. 138 283 1500 25.6 0.306 288 237.5 33,5 2000 36.4 0.215 289 2200 29.2 0.268 292 2400 22.2 0.353

64 UNIVERSITY OF MICHIGAN Engineering Research Institute IV Qualitative Observations on the Operating Characteristics of the VibratingType of Atomizing Nozzle. For each spring tension, there exists a "critical" pressure below which audible vibration will not occur. This "critical" pressure usually occurs at pressures of 50 to 100 psi above the minimum spraying pressure. As the spraying pressure is increased above the "critical" pressure, both the intensity and frequency of vibration increase. At some still higher pressure, depending on the spring tension and possibly the other design variables, two separate frequencies were noted, the new frequency being lower than the one obtained at slightly lower pressures. In some cases vibration would alternate between the two frequencies, part of the run at one frequency, the rest at the other. Generally, however, the two frequencies seemed to fluctuate quite rapidly, producing what can best be described as a "motorboat" effect. This phenomena was generally noted over a pressure range of from 100 to 300 psi. At higher pressures a single frequency, which was more closely related to the lower frequency of the dual-frequency range ("motorboat" effect) than to the higher frequency at which this dual-frequency range was approached from lower pressures, was again obtained. As the pressure was further increased, the frequency of vibration continued to increase. Qualitatively, the vibrating characteristics of the nozzle are represented in Figure 26. Z 0 u>' z DUAL FREQUENCY: I RANGE o I W | /CRITICAL I I L PRESSURE I | SPRAYING PRESSURE VIBRATING CHARACTERISTICS OF NOZZLE NO.17 Figure 26.

65 UNIVERSITY OF MICHIGAN Engineering Research Institute The two pitches (frequencies) in the dual-frequency range were usually not more than a full musical tone or two apart. However, in runs made with the smaller stem sizes (larger clearances between stem and orifice) sudden variations in pitch of as much as half an octave were noted. In general, the range of audible frequencies observed varied from a minimum of 200 cycles per second to a maximum of 800 cycles per second. Indications are that, at the same pressure, the frequency is practically independent of spring tension, but that a higher frequency is obtained with a smaller stem size (larger clearance between stem and orifice). Whenever the vibrations of the nozzle stem and spring were clearly audible, a full-cone spray was produced. For the cases in which no vibrations were audible, a full-cone spray could be generally produced by tapping the nozzle body. (See previous discussion on page 45.) Therefore, even though no vibrations were audible in the range of pressures between the minimum spraying pressure and the critical pressure, it is still very probable that vibrations below the audible range were occurring. This possibility is further borne out by the observation that the intensity of vibration increased with pressure. At the time of writing, electronic equipment for quantitatively detecting these vibrations was under construction. In general, the finest sprays were obtained at lower pressures, normally in the neighborhood of the critical pressure. At pressures above the critical pressure, the head of the valve stem was forced away from its seat on the orifice, and it appeared to vibrate in the fluid stream without ever making contact with the orifice. In addition, over the entire range of operation, the vibratingtype nozzle appeared to produce more uniform particle sizes than any of the commercial nozzles. (This fact was partially confirmed by several light transmission runs of the type discussed in Part B of this report. The rate of settling was practically linear with time, and these "decay" curves did not approach the time axis asymptotically but appeared to intersect it at a finite time.) V Conclusions. Although much work remains to be done, considerable information has been obtained on the operating characteristics of the vibrating-type atomizing nozzle. Present indications are that the behavior of this nozzle can be placed on a sound, theoretical basis. The following conclusions appear to be justified: (1) The conversion of pressure energy to velocity energy is very high as is evidenced by a coefficient of discharge of greater than 0.90. (2) For a given nozzle and stem, the aipacity and degree of dispersion can be made practically independent of the spraying pressure (obviously above the minimum) by proper selection of springs and initial spring tensions, or they can be made to vary at will by proper selection of the operating conditions. These factors demonstrate the wide versatility of this type of nozzle. (3) In the neighborhood of the critical pressure, this type of nozzle appears to produce a finer dispersion than any other commercial nozzle of comparable capacity when operated at the same pressure. Unlike other nozzles, the

66 UNIVES OF MICHIGAN 3 lUU U 22UNIVERSITY OF MICHIGAN 3 905 03525 1522 Engineering Research Institute degree of dispersion obtained from the vibrating-type nozzle does not increase wi pressure but appears to reach a maximum near the critical pressure, (4) In all cases thus far, the vibrating-type nozzle appears to produce a more uniform dispersion than any of the caomercial models tested. VI Acknowledgment. Thanks are due the Meteorological Branch of the Signal Corps for their excellent cooperation and support throughout the entire program. The invaluable assistance of graduate students, Willie Fong, L. J. Garrj son, and A. J. Stock in obtaining and correlating some of the data presented hereJ is very greatly appreciated. These men were not part of the project personnel but participated as students in a special research problems' course in the Department of Chemical Engineering at the University of Michigan. The interest of Mr. H. E. Norton, of the Army Chemical Center, Edgewood, Maryland, is appreciated~