THE iUN I V.E R S I T Y OF MI C H I G AN COLLEGE OF ENGINEERING Department of Chemical and Metallurgical Engineering.Department of Mechanical Engineering First Quarterly Progress Report INVESTIGATION OF LIQUID METAL B3OILING HEAT'TRANSFER Richard E T3alzhiser, Project Director John A. Clark C. Phillip Colver Edward E. Hucke Thomas McSweeney Herman Merte, Jr. Lowell R. Smith ORA. Project 04526 under contract with: PLIGHT ACCESSORIES LABORATORY AERONAUTICAL SYSTENS DIVISION AIR FORCE SYSTENE COMMAND UNITED,STATES AIR FORCE WRIGHT-PATTERSON AIR FORCE BASE, OHIO CONTRACT NO. AF 33(616)-8277-ITEM IIa administered through: OFFICE OF RESEARCH ADMMISTRATION ANN ARBOR January 1962 "To expedite dissemination of information, this report is being forwarded for your information prior to review and approval by the ASD project officer and is, therefore, subject to change. Any comments which you may have should be forwarded to ASD (Lt. LLoyd Hedgepeth), Wright-Patterson AFB, Ohio, within 15 days of receipt to insure correction of errors before final approval is given."

.u 1 el =. - tAAn 7 UHw I/?t) sI. "i1.V/'

FOREWORD This report summarizes efforts to date on Phase II of Air Force Contract AF 33(616)-8277. A literature survey of liquid metal boiling technology has been released as an ASD report (ASD Tech. Rept. 61-594). Phase II consists of analytical and experimental studies with boiling metal systems. The investigation is being conducted at the University of Michigan in the Liquid Metals Laboratory of the Departments of Chemical and Metallurgical Engineering and in the Heat Transfer and Thermodynamics Laboratory of the Department of Mechanical Engineering. Professors Balzhiser, Hucke and Katz of the Chemical Engineering faculty and Professors Clark and Merte of Mechanical Engineering are participating in the program. Misters Colver, Smith, Barry, McSweeney and Harrington, all graduate students in the College of Engineering, have responsibilities for specific segments of the study. Lt. Lloyd Hedgepeth and Mr. Kenneth Hopkins are serving as project engineers for ASD. Cornments are solicited by both the authors and ASD..iii

ABSTRACT An analytical and experimental study relating to the boiling of liquid metals has been initiated at the University of Michigan under contract with ASD. The program will include pool boiling studies of potassium, sodium and rubidium at temperatures up to 2200~F. Mercury will be pool boiled at temperatures near its normal boiling point as a preliminary step in studying the effect of radial accelerations up to 20 times normal gravity on the boiling process. A forced circulation loop is being designed to determine the effect of velocity and quality on -the heat transfer process. Two-phase flow studies will also be made with this equipment. iv

T A B L E 0 F.C ON T.E N T S Page INTRODUCTION 1 POTOL. BOILIXNG- STUDIES 4 FILM BOIIG SG:.:TUDIES 11 FORCED CIRCULATION STUDIES 12 iEFACIAL FFTECTS IN. LIQUID METAL. BOILING 15 LIQTID MET.AL BOIILING IN/AGRAVIC FIELDS 17 TWO-PHASE.PRESSURE DROP AND VOID FRACTION' 21 REFERENCES 25

INTRODUCTION The literature survey on liquid metal boiling heat transfer conducted as Phase I of this contract confirmed the need for reliable experimental studies in this area. The proposed use of liquid metals as working fluid.s in space turbines has raised numerous questions with regard to the effect of pressure, velocity, subcooling, quality and acceleration on the maximum obtainable heat flux from a surface to the metallic fluid. The effect of -,any of:' these variables have y-et to be substantiated. for r1ela-Gtively si.mn?le systems such as the water-steam system. With few exceptiors experimental programs to date have been confined to heat fluxes below 500,000 Btu/(hr)(sq ft)o Suggested correlations for the criticiL heat flux for liquid metals estimate the t;hese fluxes: to be in the neighborhood of 500,000 Btu/(hr)(sq ft). However, several investigators in preliminary experiments have measured fluxes substantially above those suggested.by these theoretical predictions. As reported in our literature survey, the highest flux reported in the literature to date was 600,000 Btu/(hr)(sq ft) for mercury boiling from an electrically heated surface. No effort was made to electrically insulate the fluid from the surface. Consequently, a substantial portion of the electrical energy.was dissipated directly in the mercury. This made it difficult to estimate accurately the heat flux across the mercury solid interface. It is important to note that at this estimated flux the investigators concluded that they had not yet reached e the critical flux for their particular system. Other studies including those of Lyon and Bonilla in this country and those summarized by Kutateladze (1) in the Soviet Union have investigated the pool boiling of liquid metal media. Although the effect of pressure

on the heat transfer process has been investigated by several of these investigators, none of the investigations have achieved fluxes approaching that of burnout. Likewise, none of thes:e investigations have shed any light on the effect of the other significant parameters. This program has among its goals an experimental study of boiling phenomena at fluxes in the neighborhood of 1,000,000 Btu/'(hr)(sq ft). Considerations have been given to the possibilities of attaining fluxes up to 10,000,000 Btu/(hr)(sq ft) in the event that the critical flux is found to be above a million. Both pool boiling and forced convection studies have been proposed in an effort to isolate the effects of the important parameters in liquid metal boiling heat transfer. Another objective of this program is to explore the behavior of metallic fluids in the stable film boiling regime. Some evidence seems to exist that coefficients considerably in excess of those predicted by Bromley-type correlations might be experienced. A. possible explanation for such behavior is presently under consideration. It is hoped that experimental studies can be carried out in this regime. A.-discussion of apparatus capable of operating with highly reactive fluids at the temperatures necessary to sustain film boi.ling is included in this report. Considerable time has been spent on analyzing the effect of interfacial energies on the boiling process. Uncertainty exists among investigators in. this field as to the significance of surface effects on the critical heat flux, It has been demonstrated that the. addition of certain elements to mercury systems improves remarkably the ability of mercury to transfer heat from a surface. In the case of alkali metals which wet most surfaces extremely well, it is questionable as to whether additional improvement might:be affected.by altering the interfacial energies. Certainly from.an

academic point studies in this area would be most worthwhile in providing a clearer picture of the surface phenomena involved in the boiling process. It is hoped that this program might include an experimental study of interfacial effects. The importance of accelerations normal to the heat transfer surface has been demonstrated by several investigators in the recent literature. For aqueous- systems it has been shown that the critical flux is appreciably increased by producing force fields normal to the surface in excess Of those ordinarily experienced from gravityo The desirability of such artificially induced accelerations for equipment operating under zero gravity conditions seems obvious from other points of view also. As a phase of the contract a program is being conducted in which mercury will be pool boiled from a surface subjected to radial accelerations up to twenty times that of normal gravity. The details of this program are discussed later in this report. The two-phase problems associated with the operation of such a cycle require much attention. Existing two-phase flow correlations are highly empirical and have been substantiated only for steam-water, air-water and other similar systems. The literature reviewed indicated a complete absence of two-phase flow studies pertaining to metallic media. The experimental dif icuti-s in this field are complicated by the temperatures at which one must operate and the fact that direct observation of these systems is frequently impossible. Consequently, the determination of flow regimes and the transitions between various flow regimes remains a most challenging task. It is hoped, however, that this study will produce some reliable two-phase pressure drop data which might be used to test the

reliability of existing correlations. These experimental studies are being conducted by graduate student researchers under the supervision of faculty members in the Departments of Chemical and Metallurgical Engineering and Mechanical Engineering. Several of the more challenging topics are serving as doctoral theses problems for some of these students. POOL BOILING STUDIES Heat transfer systems utilizing boiling liquid metals provide one method for efficiently exchanging large quantities of heat at relatively high temperature levelso In order to determine useful design information for such systems, it is necessary to simulate the actual conditions with a relatively simple and yet similar experiment. One might hope ultimately to generate an analytical or semi-empirical relationship substantiated for experimentally- selected'mietallic (possibly aque'iols ) sy,stems which would provide reliable engineering design- information, Unfortunately, to date no relationship exists which reliably predicts the heat transfer coefficient or the critical heat flux for liquid metals. Some experimental data for pool boiling mercury, sodium, NaK and cadmium and for mercury in a natural convection loop have been reported in the literature. For the most part, the data have not correlated well, and the highest heat flux reached in each case gave no indicatlon of having reached the critical heat flux. Ahalytical expressions for predicting critical fluxes such as the Zuber-Tribus equation have not been vexfied for liquid metals~ Conversations with persons actively engaged in boiling liquid metals at high heat f:luxes indicate considerabDle uncertainty among

the existing data. It thus seems desirable to study further liquid metals at higher heat fluxes with the hope of reaching the critical flux. In.order to operate a boiling system at these desired heat flux levels without fear of burnout, safeguards must be incorporated which prevent surface temperature excursions when the critical flux is reached. In accordance with requests from the Aeronautical Systems Division, potassium will be studied initially at temperatures up to 2200~F, Sodium and rubidium will be investigated in subsequent runs. Proposed Experimental Design The present experimental design consists primarily of a small columbium boiling vessel containing an electrically heated tube, a water-cooled or air-cooled condenser, an outer protective vessel, auxiliary and guard heaters and necessary instrumentation. Figures (1) and (2) show two alternate designs of the experimental system. Upon inspection of each figure, it is seen that the principal difference in each design is the method and type of heat transfer media used for condensing the liquid metal vapor. The design in Figure (1) uses air as the condensing fluid while the scheme shown in Figure () utilizes a conventional-type water condenser, It should be pointed out that water in contact with potassium (particularly at the high temperatures involved here) creates a violent reactiono For this reason, it is imperative that any water used in the apparatus be well confined and far removed from the boiling metal, In either design.boiling takesI place- in a columbiumn vessel 3-ino I.D. 4 inches long with approximately 1/8-in. wall thiekness. A small composite tube, exstending through the vessel, is to serve as the boiling surface. This tube is shown in Figure (3) and consists of a 0,25-in. OD.,

0.01-in. thick, and approximately 405-in. long tantalum tube, plasmasprayed with pure aluminrum oxide to a layer thickness of 0.005-in. The alumina powder to be used for the lay-er has a grain size <325 mesh yielding a minimal porosity in the coating. Over the alumina a columbium coating will be plasma-sprayed to a thickness of 0.01 to 0.04- inches along the boiling length. On the region of the tube that coincides with the vessel wall, a build-up of sprayed columbium will be allowed (0.125 to 0.25 in.) such that the composite tube can be welded directly to the boiling vessel. As indicated in the diagram, each end of the composite tube will be welded to type 316 stainless steel tubing. This tube in turn will be directed out of the protective outer vessel and connected to the helium system. The two Pt-Pt, 10% Rh thermocouples (shown on the drawing) will be used to measure temperatures that can be used to determine surface temperatures. -hese thermocouples are to be fabricated by Motech Company of Kansas City, Missouri, and consist of a core metal. ~'iLi- imilar annular metal. The metals are electrically insulated from each other by a 0.0005-in. thick layer of oxide. This coaxial assembly will be pressfitted into the inner tube and will protrude from the surface of the tube so that after spraying the oxide coating on the tube the thermocouple may be ground off flush wi th the surface of the oxide. The response time of such a thermocouple is in the order of 0.3 milli-second for an instantaneous temperature change of 1300~F. Power to the boiling tube is to be furnished through two bus-bars (type as yet undetermined, but will probably be water-cooled), clamped or welded to the inner tube and set directly against the outside wall of the boiling vessel. Total power to the boiling -tube will be determined from voltage and amperage measurements made directly across the tube. The

columbium vessel will be electrically insulated from the bus-bars by a thin layer of plasma-sprayed oxide, sprayed over the vessel. Heavy leads welded to each bus-bar,pass through the outer container wall and are connected to a DC power supply. This power source will supply 40 KW with a rated output of 1700 amps at 24 volts DC. At certain times during the investigation it may be necessary, as a precaution against burnout, to flow a stabilizer gas (helium) through the core of the heater tube. The gas will be furnished by cylinders at 2200 psi and will pass through a purification train to remove essentially all oxygen, flow through the heater tube and exhaust to the atmosphere at high velocity. As much as 300,000 Btu/(hr)(sq ft) (based on the boiling tube O.D. ) will be transferred to the gas as it flows through the tube. The cooling effec-t of this gas will control the temperature of the heater tube and thus prevent burnout at total power inputs above those required to produce the critical heat flux. The inlet and outlet temperatures and the flow rate of helium will be instrumented. From these readings, the amount of heat transferred to the helium can be calculated. The helium purification train is to be fabricated from stainless steel and designed chiefly to remove oxygen, The gas heaving the storage cylinder will be pressure regulated and then pass through five metal con. tainers (2-in. I.D. by 10 inches long) containing either hot liquid sodium or NaK, Each container in the train is to be heated by nichrome wrapped heaters, varaic controlled. Insulation will cover the complete arrangement. The entire boiling vessel aside from the condenser is to be contained in a stainless steel vessel 18 inches I,D. and approximately 1/2-in. thick. Thi.s vessel is nearly identic,la] in each design and consists of a flanged

top and bowl shaped bottom. The flange is to be gasketed and will use stove bolts and nuts to assume a leakproof fit. It. is seen that all entry into the vessel, save the guard heater leads, is made through the vessel top. This fact permits easy disassembling of the apparatus for internal inspection or repair. The inner space -between the columbium boiling vessel and the outer stainless steel vessel will be evacuated and refilled with purified helium. At all times the pressure of the helium in the inner space will be equalized with the boiling vessel pressure. In addition, the inner space is to be filled with insulating material and as a further means of preventing heat loss, a nichrome wire guard heater is to be installed as shown, Power to this heater will be regulated by a v-araic connected to a.',110 v AC circuit.'The air-cooled condensers in the first design (Figure 1) are wel6led to the top of the boiling vessel, Each condenser tube will be made from columbium and will be fabricated by Wolverine Tube Company of Detroit, Michigano They will be Wolverine Trufin type, 3/4-in. O.D. (19 fins/in. ) tubes with an average wall thickness of 0.082 inches. Assuming an airside heat transfer coefficient of 10 Btu/(hr)(sq ft).(~F) (this coefficient corresponds to an air velocity of 15 ft/sec), it was found that 6.48 total feet of condenser is needed to operate the system at 2,000,000 Btu/(hr) (sq ft)o Five 1.3 ft. condenser tubes extending from the top of the boiler satisfies this requirement.. Condenser tube velocities, using total boil-.up rate, are 132 ft/sec for one condenser tube to 26 ft/sec for five con-densers 3Po guard agaiast oxidation, the condenser and exposed boiling vessel loop will be coated ith an oxidation resistant materialo Inspection oof the literature shows that nickel-chromium alloys, eOg. 8~0 Ni-20% Cr, have

goo oidation resistance t properties in:.the neighborh ood of 2000~0F.' Work done atBat-telle Memorial TIhstitute:on zixnc coatings on columbium indi:f - cates that, a 6 rmil coating of "zinc on columbium has a service life of 1000 hours at 1800F and 70..hours at 2.000. oUnf ortunatel i't as found that although zinc- was attr.aettive in t$he: range 1600-.2OO/)0F' that at...lower temperatur.es., say 1400~F,, the zinc coating inconsistencies, pin-holes 5 etc.. were. not self-.heali:ng:as at the igher temperature and thus did. not gi.ve a sat isfa-ctory protecti'e.. coat0 Air veloci:tie.s across the ondenser section will be ufurnshed by a Z4'ge capacity blower. A duct.ill be installed so that the air:can..be for-eed directly across the ondenser bneo atch condenser tube is.. to be elded it;o a comton heer mie from standard 3 —ino 0oD. type 316 stainess steel pipe, fitted wi:th a stainless steel -top adt bottom. A. coiled water cooler will. be. used arouxnd the outside of t;he'header. A standard 3/4-in., O.D. type 31.6 stai:nless steel tuabe welded to the top of'the header allows entry int-o the system.'Througth this line, helium., thermocouple. lea.ds and: a connected vacuum -line;, enter the apparatus~ Also. beyond -the header a.nkock-cout- drum.wl.l be.installed to further pre.vent sma.ll amounts of potass s.ium Vap or from passing upstream' through t;he, heliu. m.lin.es..The alter~ate design.(,..gue 2)) us.es'.water as the condenser medium., From a.4-iin. diameter 316 stainless steel cylinder, 5-3/4-1in. diameter holes are drilled as sthown (Figure.' 2.). Four groo'ves are:Lmachined. along;the length of the cylinder, stopping approximately two inehes from either end, An anular plate welded to the condenser cylinder and a 5k-in, O.:i. shell forms the water passageo Two 3/4-ino, diameter water taps are installed to the bottom and the top o:f the.condenser shell, ColUmbium

10 tubes (3/4-in O.D.o ) are to be press-fitted down through each of the five holes drilled into the cylindero In turn, these tubes are press-fitted and welded into coinciding holes drilled in the top of the columbium vessel. The top of the condenser section is welded to a 51-in. I.D. header, possibly made from an extension of the outer water shell. This header, as well as the condenser shell, is to be made from type 316 stainless steel, Design calculations have been made to determine the required condenser lengtho It has been found that for a boiling heat flux of 10,000,000 Btu/(hr)(sq ft) and a total temperature drop of 1300~F that 0.73 ft. of condenser is needed, Using this as a basis, it is probable that a two foot condenser will be used. Instrunmentation in either system is to include a multi-point temperature recorder for indicating many necessary temperatures throughout the system. In addition, a potentiometer will be used to measure temperatures where accuracy or precision is required. Experimental Procedure The system is designed such that the experimental procedure will be simple and direct, After performing the necessary preliminaries before each start-up, the power input to the boiler tube will be set at the desired heat flux level. When the boiling process becomes stable and the system reaches steady state, the required temperature., etco readings will be made and recorded, Subsequent runs at other heat fluxes will require only adjusting the power input and waiting until the system reaches steady state, Results Anticipated

It is hoped that the following results will be obtained: 1. The determination of the boiling curve for potassium at atmospheric pressure up to and including the critical heat flux. 2, The determination of the boiling curve for potassium at other pressures in the range 0-200 psia. FILM BOILING STUDIES Operation in the transitional and film boiling regimes is theoretically possible with the pool boiler discussed earlier providing it is capable of reaching the critical fluxo In the event the critical flux exceeds that obtainable, it is impossible to raise the surface temperature to a sufficiently.high level to sustain a stable vapor film. However, if one first obtains a surface temperature above that required for stable film boiling and then contacts it with a saturated liquid metal, stable operation might be achieved.at the fluxes permissible, Consideration has been given to the design of a- submersible ribbon element which could be adapted to these conditions, Direct resistance heating may be possible as the vapor blanket, if achieved. could electrically insulate the surface from the fluidd Condensing metals might also be used in a similar manner..to study film boiling. Surface temperatures sufficiently high would first be obtained after which the metallic fluid will be introduced. In the case of an electrically heated ribbon, the power input could be adjusted simultaneously to the immersion to a value characteristic of film boiling. Further study in this area will be carried. out during the next quarter.

12 FORCED CIRCULATION S-STDIES Design studies have beenr carried out for a forced circulation- loop to study the effect of velocity, quality and subcooling on the boiling process o Based on these studies, the following specifications have been set as operating objectives flor boiling -potassium: Fluid flow rate, potassium 0.2 to 2.0 It% Maximum heat input, preheat section - 50 Kw.Maximum heat input, 5-in. test section - 15 Kw Operating fluid temperatures - 1400-1800~F Saturation pressures - 14-7 - 81 psia Tubing - 5/8 inch O.D. Haynes 25 alloy Under these specifications, liquid velocities will range from 0.3 to 3.3 ft/sec affording Reynold's numbers of 6500 to 65,000. Vapor qualities of up-to 100% can be attained at 0.7 gpm. Vapor velocities will range up to 600 ft/sec affording Reynold's numbers up to 200,000 although critical velocities, known to be much lower in two-phase mixtures,:will likely limit the operation to much lower vapor Reynolds numbers. Energy balances aound the preheater and test sectioris will be used to estimate inlet and outlet qualities and the heat flux to the test section, Surface temperature measurements pose a difficult problem. Various possibilities exist for each proposed test section. For a knowvn flux a single temperature measurement at a knownm position in the wall can be used to estimate the surface temperatureo, Several such measurements will be made to confirm this value, A specially instrumented section is planned between the preheater and test section for studtying two-phase flow behavior, This is discussed later in the report.

13 A firm proposal has been received from one prospective supplier whiAie a second is still in the process of making a design and cost estimate. Preparations are also being made to design and fabricate the equipment ourselves if proposed costs are prohibitive. it is anticipated. that potassiumwill be pumped with an electromagne-tic pump,.metered. by an electromagnetic flo.wmeter and preheated to vapor qualities of up to 100% by a three-phase direct resistance connection to the pipe. The fluid then enters the test section where fluxes of up to one million are anticipated., The stream may then pass directly to a con —.denser-subcooler or to a liquid-vapor separator. In. the latter case the liquid portion is metered for a di.rect measurement of quality and then flows..to a. hot. well. The vapor.-stream. is condensed i n a -finned-tube,. aircooled.condenser and then is added to the liquid in the hot well, Looppressure is controlled by adjusting, the rate of heat removal in the con-.denser. From the hot well, the fluid would.then -enter a subcooler where the stream temperature is lowered.to 1400~F before entering the suction side of the electromagnetic pump. Embrittlement problems with Haynes 25 require that fluid temperatures be kept above 1400~Fo Stability will be enhanced by the presence of- a throttling valve located downstream from the flowmeter. The most critical component of the loop is the test section where extremely high heat fluxes are desired. In our design we have considered several heating possibilities, of which onlyr the first two listed below seem feasible. A conceptual design is shown in Figure (7.). 1, Condensing of Liguid Metals If condensing sodium' (or lithium) is used.on the outer wall, the temperature drop between the condensing and boili.ng liquids will likely

14 be around 200~F. However, containment of sodium (or lithium) vapor at 2000~F is a definite problem. CondenSing coefficients have not been well substantiated under these conditions so some uncertainty exists here also.'This type of heating would permit operation near the critical with little or no risk of burnout should this condition be reached. 2. Resistance Heating In this method a sand.wich technique is used to heat a thin layer of refractory metal which is electrically insulated from the base tube by another -thin layer of a metallic oxide. These materials can be flame or plasma-sprayed to thicknesses of a few milso A typical sandwich would be a base tube of columbium covered with 5 mils of beryllium oxide (chosen for its high thermal conductivity), and 3 mlls of tantalum. Approximately 1000 amps will pass through the outer layer. A test program is planned to determine the limitations of this approach. Investigators at General Electric have already established its feasibility. 3. Induction Heating A flux of 108 Btu/(hr)(sq ft) can be easily obtained with highfrequency induction heating. However, data obtained are likely to be inaccurate because of the difficulty of measurng power input and of preventing fringing electromagnetic fields, 4. Radiant Heating A temperature of 50-00F would be required to transfer 106 Btu/(hr)(sq ft) to the tube walls by radiatlon. This is in excess of normal operating limits for the radiator, 5. Gas -Fired Heating Et is virtually impossible t;o obtain the required. heat flu;x via gas

15 fired heating, Another problem of considerable importance is the pressure drop likely to be encountered in two-phase flow. Based on evaluation of the data for two-phase water flow, we may expect a pressure drop as high as 12 psi/ft. (for potassium at 1700~F, 2 gpm flow rate and 50% quality). It will thus be necessary to make all lines as short as possible and to design the condenser and pteheater for minimum pressure drop. INTERFACIAL EFFECTS IN LIQUID METAL BOILING Almost all research on liquid metal heat transfer has attempted to fit existing heat transfer correlations. In most heat transfer correlations, dimensionless groups are used. This means these curve fittings lump many parameters together and measure their overall effect. In order to get a better idea of the difference between ordinary liquids and liquid metals, the variation of the parameters which make up the dimensionless groups must be studied. The experimental procedure used in liquid metal heat transfer studies must be different because of the impossibility of making: visual observationso The boiling site in a liquid metal can not be seen so it is seemingly impossible to measure any variables on the boiling surface where neither the number of boiling sites nor their position can be directly observed. A practical solution seems to rest in the measurement of the effect of variables on the dynamics of a single active site. If a hot spot or "point boiler" were used, the position of the active bubbling site is known. Also since this is the only point from which boilingis taking place, the variation of surface properties over the boiling:area will be small0

16 One of the most elusive variables to characterize has been the surface. On a normal "clean" surface it has been shown that there are many layers of foreign material. Since these layers are difficult to characterize and may continually change, it is difficult to obtain reproducable results for experiments depending heavily on surface properties. By ion bombardment cleaning it becomes possible to remove almost all of the surface contaminants. This should make the surface much more reproducable. Once the surface is clean it becomes possible to characterize boiling from a true liquid metal-metal interface. It has been shown that both electrical and thermal energy are transferred across metallic interfaces primarily by electrons. The relation between the electrical and thermal conductiyities is given approximately by the following equation (Ref. 2): K = 2(h)2 (1) K - Interfacial thermal conductivity a' -.Interfacial electrical conductivity T - Absolute temperature h - Planck's Constant e - Charge on an electron Although in-estigators have tried to verify this relation for metallic interfaces, the results have not yet been reproducable. In general it was found that the group KzT varied only slightly with temperature but changed with time, This scatter would appear to result from surface pheinprmena due to failure in achieving perfect electronic conductivity across the interface,.It is hoped. that by ion bombardnrteitb cleaning of the surface a more int;imte contact can be established. between the phases, If the surface is mace small, a large variation in the temperature of the heating element can be expected as bubbles formn n the surface XOn

17 a large boiling:surface these same temperature variations have been observed. However, it is impossible to know exactly where the bubbling site is in relation to the thermocouple. In the case of a single bubbling site the temperature measured will be directly below the bubble. Therefore, the temperature variation will take on more significance. In addition to these variables, the frequency of bubble formation and perhaps the bubble size might be. determined, by noting changes in interfacial resistivity. Also, it would be quite easy to study the effect of additives to the liquid as well as the solid. A limitation of using the "point" boiler is that the data from one active site probably will not correspond to the data from a surface with many active sites. However, a change in any boiling characteristic of the point boiling system should correspond to a similar change in the boiling characteristic of a large boiling surface. Also, the effect of impurities might have a much larger effect on a single bubble site than on a large boiling surface. It is felt, however, that this limitation can be overcome if a certain amount of care is exercised. At the present time the /advan-tages- of using a "point" boiler seem to greatly outweigh its disadvantages when considering the amount of information that can be learned. Active consideration is now being given to the design and instrumentation of the apparatus. LIQUID METAL T BOILI;NG IN AGRAVIC FIELTDS Test Apparatus A preliminary sketch of thke test; vessel to be used for studying the pool boiling characteristics of mercury in high force fields is shown in

Figure (4),, Slight modifications may be necessary as further design details are worked out, The preliminary measurements to be made are heat flux, heater surface temperature, liquid temperature, and dimensionless total acceleration a/g,, Using an electrical resistive heat source it is expected that the heater surface temperature will be the only dependent variable, besides the hydrostatic pressure at the heating surface. Initially, efforts will be made to obtain data with saturated liquid. A controllable degree of subcooling will depend upon the thermal interaction between the condenser coils and the mercury vapor or liquid. To obtain data with large subcooling it may be necessary to design the condenser coils such that they are in direct contact with the liquid. Tests using the proposed vessel will be operated at atmospheric pressure only. A test vessel for operation at higher pressures would become too massive for use in the centrifuXge presently, available. The test vessel is desi,-ed to provide a flat heat transfer surface 2 inches inr diameter, Thne liquid metal to be used initially will be mercury. Should the results warremt, consideration can then be given to other liquid metals. The test vessel is being designed for an arbitrary maximum heat flux of 500,3000 Btu/(hr)(sq ft). The value of the peak heat flux of mercury at atmospheric pressure is uncertain, but a review of the available data (Ref. 3) appears to indicate a probable value somewhat higher than this. Operation near the peak. heat fluxs is undesirable vwith this system because of the possibility of burnout. The heat souce consists of 7 elec-trical coartridge type heasters embedded in the end of a cy1lind-rica -copper block. Copper is used to minimuze the temperature difcerence between-the cartridge he-ater and the

19 boiling surface. The heaters are spaced such that each is surrounded by an equal volume of copper, and the larger diameter of the cylinder permits a lower operating heat flux in this area. A transition section to the smaller diameter of the heat transfer surface is provided along with 1-iV inch long smoothing section to assure a uniform temperature at the heat transfer surface. Measurements are planned to check the uniformity of the surface temperature, The copper heater block is isolated from the mercury at the boiling surface by a stainless steel foil 0.001 inches in thickness silver-brazed to the copper to provide intimate thermal contact. Extension of the foil from the heater surface furnishes a convenient seal between the heater and the container side-wallo A thin foil is desirable for two reasons: the temperature drop across the foil is minimized, as is the heat loss from the edge of the heater due to the fin effect. A smaller temperature drop across the foil will permit a more accurate evaluation of the heater surface temperature, as the point nearest the boiling surface accessible to a thermocouple will be the copper-stainless steel interface. The surface temperature will be calculated from measurements at 3 locations at the copper-stainless steel interface. Two thermocouples are planned for the measurement of the mercury temperature at two variable depths. The therniocouples to be used are chromel-constantan, selected because of their stability and high thermoelectric power. These will be -encased in stainless steel tubing and calibrated at the steam point, tin point, mercury vapor point and sulphur point. The emf measurements will be made with a Leeds and tothrup TjYpe K-3 Potentiometer, using an ice point reference junctiono The maximum design heat flux of 500,000:3tu/(hr)(sq ft) corresponds to a power input of 3170 watts, The power will be measured with a

20 calibrated voltmeter and armeter, To insure that the heat losses from the bottom and sides of the heater block are kept at a minimum, guard heaters and radiation shields are provided. Since the heater block varies in temperature along its length, the guard heater is split into two sections, each being independently monitored by a differential thermocouple and controlled for a minimum temperature differential. Guard heater power is supplied by resistance strip heaters wrapped around the circumference, The wa1lls enclosing the boiling surface will be composed. of two concentric stainless steel tubes vwith insulation in the annular region and outside. The mercury will be condensed at the top of the boiling chamber by a coiled stainless steel tube through which water will flow. The boiling camer is vented to the atmosphere by means of a vent tube terminating near the center of rotation of the centrifuge to prerent the possibility of an aspiration effect aecreasing the pressure within the test vessel during rotation. The- vent tube Will be water cooled for a considerable lengtg' to prevent mercury vapor from escaping to the atmosphere. TInitial test.s will be conducted with mercury depth constant at as low a value as is practical, perhaps ~ to 1 inche to keep the changes in hydrostatic pressure at the heater surface due to varying acceleration small, A question exists as to whether it is possible to attain saturated mercury at the boiling surface-with finite depth. With mercuty boiling at standard gravity; it was found that the bulk mercury temperature was essentially that of saturated mercury at the liquid-vapor interface (:ef.' I), hence somewhat subcooled at the heating surface. With high force fields this may also be the case, since the fluid motion between the heater surface and liquid-vapor' interface will be greatly

21 enhanced. A similar effect has been observed in high g work with water (Ref. 5), The only way to minimize this effect is to use very low depths of mercury, which decreases the variation in hydrostatic pressure from the heater surface to the liquid-vapor interface. TWO-PHASE PRESSURE DROP AN]D VOID FRACTION Although an extensive literature exists on the subject of two-phase fluid flow, no invrestigations haave been reported on flow of metallic systems. The mechanisms of two-phase flow behavior are not amenable to mathematical analysis without resort to highly simplified models which are frequently unrealistic. Consequently, one generally has to rely on semi-empirical correlations for predicting two-phase pressure drop. The Martinelli-Nelson method is probably the best existing method for predicting pressure drop in forced-c-irculation boiling;, and the -Lockh9art-Martinelli correlation is most often used for isothermal situations. These correlations, and others, are based on water-steam, water-air, or hydro-.carbon-air data. Thus, a need exists to experimentally examine the validity of such correlations for use in metallic two-phase flow problems. The proposed research program is aimed at answering this need in addition to providing reliable pressure drop and void fraction data for potassium flow. Hopefully, this study will reveal which parameters best characterize two-phase metallic flow. The literature is being searched for appropriate models and experimental studies which may aid in this research. It is expected that fluid physical properties play a prominent role in two-phase flow behavior, The viscosities, densities and surface tensions of sodium, potassius and their vapors were compared with those of the water-steam system on a reduced temperature, T r basis. In the

22 range 0.5 K Tr, < 0, 75 very favorable comparisons were found —i. e,, at least order-of-magnitude agreement existed in all cases. Certain dimensionless groups and comahi. nioss of physical properties entering into known important dimensionless grouaps were also plotted against reduced temperature. Again, order-of-magnitude agreement between sodium, potassium and water w as noted, as.nd indeed? remarkably good agreement exists in some cases, These fsarorable physical property comparisons suggest that the Martinelli correiation.may predict pressure drop in metallic floaw to at least the correct order of magnitude. It is suggested that the correlatLions be applied on a reduced-property basis. If the reduced temperature in a problem is in the range O05K(Tr<O.75 the numerical results are postulated to be as accurate as can be obtained without help of experimental data. In the experimental program, pressure drop measurements alone will not be sufficient for correlation and.analysis. In. addition? it is proposed to obtain void fraction data, the importance of which will be outlined here. In gas -liquid ~r!lo, the- man velocities of the two phases are not equal., This condition is _kncwnm as'tslip"t and occurs both: in horizontal anad vertical floTw. T'the true proportion of the pipe crossseetion occupied by either phase differs from that calculated, if the calculation is based on tLne volums of liquid and gas passed th.rough the duct. To compute -the true ga,s velocity., one must know- the vapor volume fraction,'The slip velocity ratio is given by vg X ( 1: ( -(2) where Vg VX 2- vapor a'nd liquid superficial velocities pgr p- = vapor and iTiquid densities - quality, mass of vapor/imss of mixture v =vapor volume fraction (also called void ~raction or liquid holdup)

23 The hydrodyrnamics of two-phase flow depend strongly on void fraction. In a flow system in which a change of phase is occurring, the pressure drop can be written as -dP static = dP frictional + dP acceleration + dP hydrostatic (3) overall- head or by the faxmiliar energy balance -dP = PmdHf + Pm g + pmdL m f PMg m (4) ewhere pm,= mixture density H, - frictional head loss V,= local superficial velocity L = length of duct g,= force-mass conversion constant For a two-phase mixture with slip, the local mixture density is given in terms. of the local void fraction. Without knowledge of a, the hydrostatic, frictisnal and acceleration components of pressure drop cannot be separated from the overall static pressure drop. From the above remarks and equations, the importance of measuring'void fraction, ac becomes evident.'The ex-perimental data will be obtained from the boiling potassium circulation loop. It is hoped that study can be made on b-oth horizontal and vertical flow, but this aspect of the research has yet to be decided. Pressure gauges on a test section of known length will furnish the pressure drop data, Gamma-ray attenuation me'thods have proven successfu for making void fraction measurements up to qualities of 10 to 13 per cent. Aborve this quality level it is difficult to obtain

24 reliable data as can be seen from Figure (5). Above x O, 1 the slip' ratios bunch together, making it very difficult to establish Vg/VI for any given ac value. In -the radiation attenuation method of. making void fraction measurements, low energy radiation is desired in order -that a large portioz of. the incidelnt beam can be attenzuated. It is proposed.to use low energy X-rays wsich will eliminate the half-life problem rassociated with. gamma-ry sourees as weLl as mnAing it possible to have a reasonably monoenergetie beam- -In receivirg and -reeording the transmitted beam,. a direct current amp'lifying g system- is.suggested. Only current measurements are re-corded in calibratio:n and _data runso Fig~ure (6) gives a schema;tic diagra;m of the proposed X-ray attenuation setup. Be-ease of its tempMra.:ure sensitivity3, provisions will be'made tO o tie scintillatinti crystaL., it ill also be neessa-ry to shield the photomrultiplier tube from. electrostatic and ele-trocmagetic fields present arouxid the loop. Concentrie sheLLs of'copper, mu-metal, and iron canibe used for this sltaelding. It will be reqxired -that the photomultiplier tube have a lw dark -urrent, In proposing.th.fis research programi, costs have been considered, he design of t-,he eirev-2at'ion loop includes instruerntation Vich will pro-ide heat balance data for determinirng quality. The Mine Safety The equipment for measnunri the ktrsmi tted X-ray be-am i void fraction determinations ill cost about $2~00 An X-ray apparatus will probably be avai3lable at litt%le or no cos-t,. athas appears that the reear program'is finacia alfi feasible,,

REFERENCES 1. Kutateladze, S. S.,, "Liquid Metal Heat Transfer Media," Supplement #2, 1958, N.Y., Consultants Bureau Inc., 1959. 2. Bonilla, C. F. and Wang, S. J., "Interfacial Thermal and Electrical Resistance between Stationary Mercury and Steel," NYO3091. 3. Romie, F. E. et al., "Heat Transfer to Boiling Mercury," Trans. ASME, J. Heat Transfer, Series C, Vol.o 82, Noa 4, Nov. 1960. 4, Bonilla, C. F., et al.,, "Pool Boiling Heat Transfer with Mercury," Reactor Heat Transfer Conference of 1956. 5. Merte, H,, Clark, J. A., "Pool Boiling in an Accelerating System," Trans. ASME, J. Heat Transfer, Series C, Vol. 83, No. 3, Aug. 1961.

II. Iii' 1K~~~~~~~~~~~~~~1 LQ r~~~~~ *,~~~~11cz~~~4j4 czW) ~~~~~~~~~~~C ~

Li -' CX )-. I. s. 1K r=i n~~~~~s II II 1sI Li Li~ ~ s

LlJ O~~ K (33 14J~~~~~~~~~t LU~~~~~~~~~~~~~~~z

z ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~I —r~ w Q~~~~~~ ZW W LL -. I ~ W LiLJ o J (I) w LL w~' z w O W — w Ir w LLL w a. WWf) z cr crcr a. crz ) CD w W =l U cn r cn - H w I.w W H CD. CD 0 H F W W wa w Ul, an Hw.WZ~N0 w CDCD ~ ~ ~~~~~C o LL X 1) c CD CDW cu C <f L D J [ C -0 0 -- -C C/ 0 H o -H I 0 -0D < W O q( Ci z w~ CD CD~ ~~~L % w C apz ar _ ocr X cr -j 3 0 ~~~~ C.. 1 a.,iw z ~' z acr ) X CL Q3Lr cn Lt, 3 e e CIJ,- CM z ccJ o ~z4 IN 41~~~~~~~~~~~~F 12.~O (._m 1 1,L) Cc tr~~~ St1, O.', 1 /i ------ 001 lo H 5 T% \,1", \ 2 z N -Ju.L) LLI1 Lj 1 X ~ ~ ~ ~ ~ ~ (II 0. 0 ui j Lio D C).a) cl

0o9 0.8 0.7 0.40.3 0.2 I' 0.1 O.I'I I 0 0. 0.2 0.3 O, 4 0.5 0.6 0.7 0.8 0.9 /.0 VAPOR OUALITY, X Fig. 5. Void fraction as a function of quality for various slip ratio pararmeters (calculated from potassium data at 17400F using Equation (2)). P;8r~rAsE Y(i)t~~ TUBE \ HIGH VOLTA6E SUPPLY FLOW CHAMiVNEL PHOTOMULT/PL/ER TUE3 AC. AMPL /F/ER SC/A/TILLAT/ON CRYSTAL PREAM,PL/FIER Fig. 6. Schematic diagram of void fraction measurement apparatus.

Fig. 7. Conceptual design of the forced-circulation liquid metal loop. 8vx CoNfmecr,oo / p T73r Se~cr oA is K v 13 us SU-COO, S,F I PVVr C.'r 0 Ala PrH Ar tV C Me,, /,,,/ F~1 I~ I _~ rvlcFir i _ p'''' w "'- ---- - It. j /, ~.,tt~'...........

UNIVERSITY OF MICHIGAN [llll I 1 011111 11 1 I2596l llI 022 2596o;;s