THE UNIVERSITY OF MI CHIGAN COLLEGE OF ENGINEERING Department of Mechanical Engineering Heat Transfer Laboratory Technical Report No. 3 INCIPIENT BOILING OF CRYOGENIC LIQUIDS Kenneth J. C'oeling Herm'r'i rte.Jr. /-......;. *', ";;...., /: (..., -:,.d r.ay I.i /: 6 -O ORA. Proje. T4. p461 under contract with: NATIONAL AERONAUTICS AND SPACE ADMINISTRATION GEORGE C. MARSHALL SPACE FLIGHT CENTER CONTRACT NO. NAS 8-20228 HUNTSVILLE, ALABAMA administered through: OFFICE OF RESEARCH ADMINISTRATION ANN ARBOR January 1968

This report was also a dissertation submitted by the first author in partial fulfillment of the requirements for the degree of Doctor of Philosophy in The University of Michigan, 1967.

TABLE OF CONIVTENTS Page LIST OF TABLES iv LIST OF FIGURES v NOMENCLATURE ix ABSTRACT xi Chapter I. INTRODUCTION AND LITERATURE SURVEY 1 A. Introduction 1 B. Literature Survey 2 II. EXPERIMENTAL APPARATUS 16 A. Dewars 16 B. Pressure Control 18 C. Surface Mounting 21 D. Support Tube 25 E. Cover Plate 25 F. Heater 26 G. Power Supply 26 H, Radiation S:hields 26 Io Convection Shield 27 III. INSTRUMENTATION 28 A. Thermocouples 28 B. Potentiometer 31 C, N'ull Detection 31 D. Pressure 32 E, Heater Voltage 32 F. Heater Current 32 IVo PROCEDURE 33 A. Surface Preparation 33 B. Assembly 37 C. Purging 38 D. Filling 39 E. Establishing Test Conditions 40 F. Testing 41 G. Vibration Tests 44 iii

TABLE OF CONTENTS (Concluded) Page H. Replacing Thermocouple Leads 44 I. Surface Measurements 45 J. Data Reduction 45 V. RESULTS 47 A. Natural Convection 47 B. Nucleate Boiling 56 C. Incipient Boiling 86 D. Theoretical Analysis 94 E. Artificial Sites 98 F. Surface Measurements 100 VI. CONCLUSIONS 107 A. Natural Convection 107 B. Nucleate Boiling 107 C. Incipient Boiling 111 D. Recommendations for Future Work 112 Appendix A. THERMOCOUPLE CALIBRATION 114 B. ERROR ANALYSIS 118 1. Heat Flux 118 2. Temperature Measurements 121 3. Position of Liquid Thermocouples 123 C. DATA 124 REFERENCES 153 iv

LIST OF TABLES Table Page I. Theoretical Marginally Active Cavities 99 II. Artificial Sites 100 III. Surface Roughness 106

LIST OF FIGURES Figure Page 1. Schematic of test apparatus. 17 2. Liquid dewars. 19 30 Ma:nifold ring and cover plate. 19 4. Schematic of test surfaces and mountings. 22 5. Test surface. 23 6. Clamping arrangement. 23 7. Housing cup. 23 8a. Mounted surface without convection shield. 24 8b. Mounted surface with convection shield. 24 9, Control panel. 29 10. Measuring instruments. 29 1. Temperature distribution in liquid nitrogen when heat downwards. 48 12. Natural convection heat transfer to liquid nitrogen. 49 13. Natural convection heat transfer to liquid hydrogen. 50 14. Hysteresis of polished stainless steel surface heating upwards in liquid hydrogen. 57 15. Hysteresis of 280 grit stainless steel surface heating upwards in liquid hydrogen. 58 16. Hysteresis of 600 grit stainless steel surface heating upwards in liquid hydrogen. 59 17. Hysteresis of Teflon surface heating upwards in liquid hydrogen. 60 vi

LIST OF FIGURES (Continued) Figure Page 18. Effect of rotating outer dewar on heat transfer to liquid hydrogen. 67 19. Reproducibility of heat transfer to liquid hydrogen from a Teflon surface. 68 20. Effect of convection shield on heat transfer to liquid hydrogen. 70 21. Effect of roughness on heat transfer to liquid hydrogen from horizontal stainless steel surfaces. 72 22. Effect of roughness on heat transfer to liquid hydrogen from vertical stainless steel surfaces. 73 23. Effect of roughness on heat transfer to liquid hydrogen from horizontal copper surfaces. 74 24. Effect of orientation on heat transfer to liquid hydrogen from a plished stainless steel surface. 77 25. Effect of orientation on heat transfer to liquid hydrogen from 600 grit stainless steel surfaces. 78 26. Effect of orientation on heat transfer to liquid hydrogen from a 280 grit stainless steel surface. 79 27. Effect of surface material on heat transfer to liquid hydrogen from smooth horizontal surfaces. 81 28. Effect of surface material on heat transfer to liquid hydrogen from rough horizontal surfaces. 82 29. Effect of surface material on heat transfer to liquid nitrogen from smooth horizontal surfaces. 83 30. Effect of liquid on heat transfer from a horizontal polished stainless steel surface. 85 31. Effect of liquid on heat transfer from a horizontal 600 grit stainless steel surface. 87 32. Initial vapor formation conditions. 88 vii

LIST OF FIGURES (Concluded) Figure Page 33. Model for analysis of the growth of a vapor nucleus. 95 34. Photomicrograph of polished copper surface. 101 35. Photomicrograph of polished stainless steel surface. 101 36, Photomicrograph of 600 grit copper surface. 102 37. Photomicrograph of 600 grit stainless steel surface. 102 38, Photomicrograph of 280 grit stainless steel surface. 103 39. Photomicrograph of Teflon surface. 103 A-1. Schematic of thermocouple calibration apparatus. 115 B-1. Model for fin heat loss calculations. 120 viii

NOMENCIATURE Units indicated are the units normally used, other units indicated locally where needed. Symbol Definition A area, cm2 cp specific heat, joules/gm~K D diameter of heat transfer surface, cm d diameter, cm or in. g gravitational acceleration, cm/sec2 or ft/sec2 Gr Grashof number, p2gPATL3/ A2 h heat transfer coefficient, watts/cm~K hfg heat of vaporization, erg/gm L length, cm Nu Nusselt'number, hL/k P pressure, dynes/cm2 or lbf/in.2 Pr Prandtl number Cpp/k 1q heat rate watts q"T heat flux, watts/cm2 r radius, cm or in. T temperature, ~K or ~F Vfg specific volume increase during vaporization, cm3/gm thermal coefficient of volume expansion, 1/~K ix

NOMENCLATURE (Concluded) Symbol Definition thickness, cm or in. 3Tcorr temperature drop from surface thermocouple to the surface, ~K viscosity, poise p density, gm/cm3 oC surface tension, ergs/cm2 Subscripts c cavity f liquid g gas 1 liquid L loss M measured s surface — saturation sat saturation sur surface v vapor

ABSTRACT The purpose of this work was to determine the heat flux and surface superheat necessary to initiate nucleate boiling, i.e., form the initial vapor, on a flat surface heating into a pool of -saturated liquid hydrogen. Some incipient boiling data were also obtained with liquid nitrogen, used to check out the system. Natural convection and nucleate boiling heat transfer data were obtained with both liquids when the incipient boiling tests were conducted. The variables investigated are surface material, surface roughness, and orientations. The surface materials used are stainless steel, copper, Teflon, and a special surface consisting of a glass fiber web covered with epoxy cement. The stainless steel and copper surfaces were tested in both a polished and roughened condition. The orientations investigated are horizontal upwards, vertical, and horizontal downwards. An instrumented test surface was placed in the cryogenic liquid and heated by a dc resistance wire heater. After steady state conditions were established, the heater power and surface superheat were measured and visual observations of the surface made. The visual observations were to determine if vapor was being formed and, if so, the pattern of the boiling. The heater power was then stepped to a new value, steady state conditions established, and new measurements and observations made. Natural convection and nucleate boiling heat transfer data from surfaces heating upwards and vertically in liquid hydrogen are presented. The range of heat fluxes is from less than 10 x 10 watts/cm2 to 1500 x l0-3 watts/cm2. For liquid nitrogen, natural convection heat transfer data for surfaces heating upwards and downwards and nucleate boiling heat transfer data for surfaces heating upwards are presented. The range of heat fluxes is from 10 x 10lO watts/cm2 to 10,000 x 10-3 watts/cm2. The surface superheat and heat flux when the initial observable vapor was formed are reported for 15 combinations of surfaces, liquids and orientations. The surface superheats at the initial vapor point range from almost 0K for an epoxy coated surface heating upwards in liquid hydrogen to over 60K for a polished stainless steel surface heating upwards in liquid nitrogen. During nucleate boiling of liquid hydrogen, the rate of heat transfer at a given surface superheat was as much as 25 times greater from a copper surface than from a stainless steel surface prepared in an identical manner. Over the range of boiling heat fluxes investigated, the rate of heat transfer from a surface at a given surface superheat was greater when heating vertically than when heating upwards. When boiling liquid hydrogen, the roughest stainless steel surface tested required a larger surface superheat than a smoother stainless steel surface for a given heat flux. The smoothest stainless steel surface required the largest surface superheat. It is postulated xi

that the liquid hydrogen wets the larger surface cavities. The roughest surface has many large cavities on it, so the wetting results in a number of potentially active sites being inactive and the surface superheat at a given heat flux increases. The surface superheat and heat flux at which the initial vapor was observed to form on a given surface in a given orientation were reproducible to within +25% of the average values. The initial vapor formation in liquid hydrogen and liquid nitrogen is primarily a function of the surface superheat and is not a strong function of orientation or heat flux. In general, the lower the surface superheat needed to form the initial vapor on a uniform surface, the lower the surface superheat at a given nucleate boiling heat flux. xii

I. INTRODUCTION AND LITERATURE SURVEY A. INTRODUCTION Boiling liquids, although used for centuries, received relatively little attention from researchers until the last several decades, When a large quantity of heat is generated in a small volume, nucleate boiling is especially attractive for removing the heat as it permits a large heat flux to be transferred from a solid surface (or nonboiling liquid) to a boiling liquid with a relatively small temperature difference between them. The development of nuclear reactors, jet engines and rocket motors has been responsible for much of the recent interest in nucleate boiling. In nucleate boiling, the liquid is in contact with the heated surface and vapor is formed. at (or very near) the surface. An understanding of the vapor formation process at a heated surface is thus necessary for an understanding of nucleate boiling. In many instances, it is also desirable to know under what conditions the very efficient heat transfer mechanism of nucleate boiling replaces the less efficient mechanism of natural convection. The increased use of cryogenic liquids in recent years creates a need for reliable data for heat transfer from a solid surface to cryogenic liquids. A critical review of the data available in the literature indicates that much of it is either questionable because potentially large heat losses and temperature differences between the surface and measuring device where neglected, or of limited practical use because of unusual test geometry or test procedures. 1

2 The cryogenic liquids can also serve as a rather severe test of correlations and theories as their physical properties are considerably different than those of the "normal" liquids usually used for comparison. The problem of dissolved gases is eliminated or minimized with certain cryogenic liquids. No gas can go into solution in liquid helium and only helium can go'into solution in liquid hydrogen. Being available in high purity, the cryogenic liquids are excellent for heat transfer studies. The incipient boiling of liquid hydrogen is of particular interest in the space program. For the long term storage of liquid hydrogen under near zero gravity conditions, it is important to know the heat flux and temperature at the container wall at which nucleate boiling will be initiated. The purpose of this work was to determine the heat flux and surface temperature necessary to initiate nucleate boiling, i.e., form the initial vapor, on a flat surface heating into a saturated pool of liquid hydrogen. Some incipient boiling data were also recorded with liquid nitrogen, while checking out the system. Natural convection and nucleate boiling heat transfer data were recorded for both liquids when the incipient boiling tests were made. The variables investigated are surface material, suface roughness and orientation. The surface materials investigated are stainless steel, copper, Teflon, and a special surface consisting of a glass fiber web coated with epoxy cement. The stainless steel and copper surfaces were tested in both a polished and roughened condition. The orientations investigated are horizontal upwards, vertical, and horizontal downwards. B. LITERATURE SURVEY The immense amount of literature available on nucleate boiling and on the

inception of boiling makes it impractical to attempt here a complete review of the literature. Instead the reader is referred to one of the many reviews that are availablee and only that literature of direct interest for the present work will be reviewed here. The formation of a vapor nucleus from the liquid phase and the conditions for the growth of a vapor nucleus have been investigated both analytically and experimentally. Bankoff5 applied nucleation theory to a superheated liquid at a solid surface and made an order of magnitude comparison between superheats observed in boiling and those theoretically necessary to form a new vapor nucleus at different locations. From these comparisons he concluded that during nucleate boiling no new nuclei form within the homogeneous liquid, on a flat surface, at a projection on the surface or in a well wetted cavity. Instead, nucleation occurs preferentially in nonwetted cavities, and it is likely that ebullition normally results from the growth of existing nucleio By microphotography, Clark, Strenge and Westwater6 found that active nucleation sites were located at surface imperfections such as pits and scratches. Howell and Siegel13 reported that photographs of water boiling from artificial sites drilled in stainless steel surfaces show a vapor nucleus remaining in the site after a bubble leaves. Analyses for the growth of a vapor nucleus are given by many authors.eog,7-13 The general procedure is to first relate the difference between the pressure of the vapor in the nucleus and the liquid pressure to the radius of curvature of the nucleus and the fluid propertieso Gibbs equation for the static equilibrium of a bubble,

4 2r is usually used for this purpose. The pressure differential is then related to the increase in the saturation temperature of the vapor in the nucleus by means of the Claperyon relation, dP fg (2) dT Tv fg This temperature increase is a liquid superheat if referred to the liquid pressure. Analyses differ owing to the assumption of various equilibrium shapes of the vapor nucleus, various criterion for determining when the vapor nucleus will grow and various temperature distributions in the liquid. The equilibrium shape of the vapor nucleus is that shape which it assumes before it begins to grow. All of the analyses assume that the vapor fills the cavity and extends into the liquid as a truncated sphere, a common case being a hemisphere. The radius of the truncated sphere and the amount of truncation depend upon the cavity radius, the angle of the cavity wall, and the liquid contact angle. For a hemispherical nucleus, the radius of the hemisphere is equal to the cavity radius. Generally the criterion for growth of the vapor nucleus is that the temperature at some critical location exceeds the saturation temperature of the vapor inside the nucleus. Corty and Foust7 formulated a general expression for the radius of curvature of the-vapor-liquid interface of a vapor nucleus in a conical cavity. Assuming that the nuclei were truncated spheres, that the surface cavities had cone angles of 1200 and that the critical temperature location is at the solid-liquid interface,

boiling heat transfer datawere used to calculate the size of the surface cavities theoretically serving as nucleation sites on two different surfaces. The cavity sizes so computed from data for three different liquids boiling on each surface agreed resonably well with each other. The analysis of Griffith and Wallis8 assumes a hemispherical nucleus and that the critical temperature location is the solid-liquid interface. The analysis was compared with data for water boiling from a polished copper surface on which 37 artificial conical sites of uniform size had been placed. Two possibilities are given for the discrepancy between the20oF superheat needed to form vapor from the sites and the 3~F predicted. It is postulated that either the nuclei are not hemispheres or else that the surface temperature in the immediate vicinity of a site is considerably below the average surface temperature, which is measured. When the liquid was uniformly superheated by a bath, rather than being heated by the solid surface, good agreement was found between theory and experiment. Here the surface served only to supply the surface cavities which acted as nucleation sites, The superheat was slowly reduced by reducing the bath temperature and the conditions when vapor was no longer generated from the surface site recorded, It should be noted that the foregoing analyses7'8 place no upper limit on the size of an active cavity. They indicate that the larger the surface cavity, the lower the superheat necessary for vapor growth, approaching zero for large cavities. It would appear to be physically reasonable that there should exist a size beyond which a surface cavity could no longer function as an active nucleation site~

6 10 Hsu made an analysis of the growth of a vapor nucleus in a surface cavity which is based on the observation by Hsu and Graham9 that as a vapor bubble leaves the surface, cool liquid from the bulk replaces the warm liquid in the thermal boundary layer. He assumed that the vapor nucleus is a truncated sphere and that the critical temperature location is the outer edge of the vapor nucleus, A limiting thickness for the thermal boundary layer, beyond which it does not grow because of turbulence in the bulk liquid, is assumed. The time that it takes for a transient conduction process in the liquid to heat the fluid at the critical location to the saturation temperature of the vapor in the nucleus is the waiting period, or dead time. between bubbleso The limiting temperature distribution resulting from the transient conduction process is assumed to be linear from the wall to the edge of the limiting thermal boundary layer. The analysis, using the limiting temperature distribution, predicts a range of cavity sizes that will be active, Very small nuclei do not grow as they have a small radius of curvature and are therefore at a high pressure and the vapor in therr have a high saturation temperature. Very large nuclei do not grow as their outer edge is located in the relative cool liquid away from the surface. If cavities of all sizes are present on the surface, the initial boiling with theoretically occur when conditions are such that the theoretical maximum active cavity size is equal to the theoretical minimum active cavity size. It is necessary to know the limiting thermal boundary layer thickness in order to use this analysis. A modification to the above analysisl0 was made by Bergles and Rohsenowo11 A linear temperature distribution in the liquid, determined by the surface temperature, the heat flux and the thermal conductivity of the liquid, was assumed

7 along with a hemispherical vapor nucleus. By using conventional correlations for the heat transfer coefficient, it is possible to predict the surface superheat at a given heat flux and then the size range of active cavities. The interface between the vapor nucleus and the liquid was considered by Han and Griffith12 to be both adiabatic and isothermal, before the vapor nucleus grows. Satisfying only the adiabatic condition, the critical temperature location is taken to be the outer edge of the nucleus. They concluded from results of potential flow theory and the analogy between isothermal lines in conduction and potential lines in fluid flow, that the critical temperature location is 3r/2 from the solid surface. Their analysis assumes that the temperature distribution in the liquid, at any given time, is linear with a gradient equal to the gradient computed at the wall for a transient conduction process in the initially uniform temperature liquid~ The temperature distribution is assumed to be limited by the maximum thermal boundary layer thickness that is present in natural convection. The results and limitations are similar to those for Hsu's theory,l0 with a smaller range of cavity sizes predicted to be active at a given surface temperature~ Howell and Siegel boiled water from polished stainless steel surfaces that contained artificial sites. Most of the surfaces contained two nominally identical sites, but some contained up to six sites of various sizes. The analysis of Griffith and Wallis8 predicts that all of the sites will be active at a lower surface superheat than was observed to be necessary~ The difference between the theory and experiment is especially large for the larger sites, The analysis of Hsu10O correctly predicts the minimum surface superheat necessary

to form vapor from a surface, if cavities of all sizes are present on the surface. The analyses of both Hsu and Han and Griffith12 predict that the smaller sites will be active at a lower surface superheat than observed to be necessary and that the larger sites will not be active at any value of surface superheat. Howell and Siegel observed that after a vapor bubble left the larger sites a vapor nucleus that extended outside the thermal boundary layer remained at the site, An analysis was developed, assuming that the vapor nucleus was an isothermal hemisphere at the temperature of the saturated vapor in it and that it would grow whenever the net heat transferred across the liquid-vapor interface is positive. The analysis predicts superheats for the growth of vapor nuclei closer to the observed values than those predicted by the analysis of Griffith and Wallis, However, the predicted values are still less than the observed values and there is no upper limit to the size of the cavities predicted to be active. This analysis could be used to predict a lower limit for the superheat necessary for a cavity to become active, It is important to note that all of the analyses considered for incipient vapor formation assume a vapor nucleus in a surface imperfection. The maximum cavity size that will be active in any given situation may be determined by the ability of the cavity to hold a vapor nucleus. The replacement of the vapor in a cavity by liquid has been examined by Bankoffo14 He concluded that when a liquid front moves over a surface cavity containing vapor, the liquid will replace the vapor in the cavity whenever the liquid contact angle, measured from inside the liquid, is less than the enclosed angle at the bottom of the cavity. If a

9 vapor front moves over a surface cavity containing liquid, the vapor will not replace the liquid whenever the sum of the liquid contact angle and the cavity angle is less than 180~o Additional data on the inception of boiling can be found in the literature. Marto and Rohsenowl5 boiled liquid sodium from a surface that contained twelve re-entrant cavities with an outside radius of 00002 in, A superheat of 48~F was required to initiate boiling from these sites, The analysis of Bergles and Rohsenow11 predicts that for thesame conditions a cavity with radius of 0,0028 in, will become active at 48Fo, Hutton and Hall16 boiled water from a chrome plated stainless steel rod on which small (maximum radius of 0.0025 in,) artificial sites had been made by acid etching, Much higher temperatures than predicted by the analyses 10 12 of Hsu and Griffith12 were needed for the sites to become active. The site sizes were well within the range predicted to be active under the conditions that they did become active. Defining the incipient boiling point as the intersection of the extensions of the natural convection and fully developed nucleate boiling curves. Merte and Clark17 indicate that for saturated water boiling in an accelerating system the incipient boiling point appears to be independent of acceleration level, Working with a cryogenic liquid, Lyon22 found that the heat flux and superheat necessary to initiate boiling in liquid helium was not reproducible, Even when the surface was not removed from the liquid, large variations were noted between runs,

10 An interesting observation concerning incipient boiling is an apparent difficulty in activating the first site.7'15'18 If the heat flux is increased from a low value, natural convection can be maintained at a heat flux which produces vigorous nucleate boiling when decreasing the heat flux from a higher value. The surface temperature under such conditions is much larger than when boiling is present at the same heat fluxo Once boiling begins, the surface temperature drops to a value characteristic of nucleate boilingo Corty and Foust7 report that the rougher the surface, the larger the superheat that could be obtained before boiling begins. The opposite trend was found by Marto and Rohsenowo15 Corty and Foust also reported that once the initial site became active, patchwise boiling occurred, i.eo. there was a clustering of active sites in some areas of the surface while other areas of the surface were void of active site, Patchwise boiling also occurred if the heat flux was decreased from a vigorous nucleate boiling level until natural convection was established, and then immediately increased again. The last sites active while decreasing the heat flux became active again while increasing the heat flux and served as centers for the patches of active sites~ Gaertnerl9 postulates that patchwise boiling does not actually occur and quotes Feller46 as saying, it is an "established fact that to the untrained eye randomness appears as regularity or tendency to cluster," It is well known that the position and slope of the nucleate boiling heat transfer curve is strongly affected by the surface.e'g.'7,15,16~20,21128,41 An increase in roughness generally decreases the surface temperature needed for a given heat transfer rate, Corty and Foust7 found that for n-pentane boiling

11 from nickel surfaces, the AT needed for a given heat transfer coefficient correlated reasonably well with the RMS roughness of the surface. A limit to the effect of increasing surface roughness is postulated. Berenson20 also boiled n-pentane from a number of different surfaces and reports that the RMS roughness is not a significant number in correlating the effect of roughness. It is also reported by Berenson that materials with higher thermal conductivity require a lower AT to transfer a given heat flux. Kurihara and Myers21 boiled a number of different liquids from copper surfaces polished with emery paper. They found that as roughness is increased, the incremental effect of an incremental increase in roughness becomes smaller. It is estimated that the limit in roughness above which the boiling heat transfer coefficient will not be changed by an increase in roughness is about 30 4in. The effect of orientation on nucleate boiling is not clear. Githinji and Saberski found that in subcooled nucleate boiling of isopropyl alcohol, a vertical heater had a larger heat transfer coefficient than a horizontal heater heating upward. The heat transfer coefficient for a horizontal heater heating downward was considerably lower than for the other two cases. Marcus and Dropkin report that for saturated pool boiling of water, the heat transfer coefficient at a low heat flux decreases as the surface is rotated from horizontal to vertical. The opposite effect was found when vigorous boiling was taking place. Using liquid hydrogen, Class, et al.,28 report that for a smooth surface there was very little difference between the boiling curves when the surface was horizontal, inclined 45~ or vertical. For the same surface roughened with emery

12 paper the heat transfer coefficient at a low heat flux decreased as it was rotated from horizontal to vertical. CI-vering the surface with a light coat of grease produced the opposite effect. In 1961 Richards, Steward and Jacobs25 collected the available heat transfer data for all cryogenic liquids and, when possible-, compared them to theories and correlations. The only natural convection data found for liquid hydrogen were froma vertical wire and no natural convection datawere found for liquid nitrogen. Much of the data presented in Ref. 25 are applicable only to the special conditions of the tests.: Two sets are considered important for the present work. MulPord and Nigon26 boiled liquid hydrogen from the: outside of a vertical copper cylinder. The test pressure was 0.73 atmospheres and the heat flux was varied from low nucleate boiling values to film boiling values After datawere recorded for the smooth surface, the surface was sand blasted to roughen it. The heat transfer data from the roughened surface were -within the scatter of the data from the smooth surface. Class, et al.,27 boiled liquid hydrogen from an electrically heated resistance. ribbon 1 in. wide, 22 in. long and 5 mils thick, -The test pressure was varied from 0.8-i to l,.'7 atmospheres fand -the heat flux from low nucleate boiling values to film boiling values. The temperatures were measured at various locations on the back side of' a 5 mil Mrlar film to which the heater ribbon was cememted. The complete report2' indicates that an estimate was made of the heat loss from the back side of the heater ribbon and of the temperature drop across the Mylar filmo Other heat losses and the temperature drop across the heater ribbon were not considered, Data are presented:for the ribbon in the

verticals 450 up, and horizontal positions. The ribbon was tested in an "as received" condition, roughened with emery paper and covered with a light coat of grease. The effect of orientation on the heat transfer data was mentioned above. In all of the positions, there was very little difference between the roughened and smooth surface at nucleate boiling heat fluxes near burnout. At a, low nruc.leaate boi.;Ling heat flux, the TaT was smaller for the roughened surface than for the smooth surface. In the vertical position, the greased surface had a smaller iT than the other two surfaces at a low heat flux and a larger AT than the other two surfaces at a large heat flux. In the other positions, the greased surface had a larger AT' than the ct;her two surfaces at all heat fluxes. Other significant pool boiling data for liquid hydrogenhave been reported since 1961, Drayer and Timmerhaus29 boiled from the outside of a vertical brass tube, The test pressure was 0.82 atmospheres and the heat flux limited to low values in the nucleate boiling range. Sherley32 boiled from a lead film deposited on a. horizontal glass slide, Datawere taken at one atmosphere pressure over the entire nucleate boiling range for both normal gravity and zero-gravity conditicns. The temperatures were measured by resistance changes of the lead film, also used as the heater, and represent an average across its thickness. Of particular interest to the present study, it was indicated that a heat flux of between 250 and 500 Btu/hr ft2 was required to initiate boiling under normal gravity conditions, The corresponding surface superheat was about 2. 5~F. Nearly the same incipient boiling results were found under zerco-gravity conditions, but they might have been different if zero-gravity times longer than the maximum of 15 sec had been available.

14 Data are presented by Graham, et al0 for liquid hydrogen boiling from an electrically heated resistance ribbon. Pressures ranged from 60 to 260 psi and accelerations from 1 to 10 g ~ The -temperatures were measured. on the hack side of the heater ribboln. In the only run that the temperature drop aCroTss the ribbon was estimated, the correction to the measured temperature, to ebtain the surface temperature, amounts to about 50% of the measured temperature. No corrections were made for heat losses. Steinle33 indicates that Tusk35 conducted tests on incipient boiling of Liquid hydrogen, Zero-gravity conditions were simulated by heating liquid hydrogen with a surface facing downward. Boiling began on a rough surface with a temperature rise of less than 0.2~F and on a very smooth surface with a, temperature rise of 6~F. 40 Hord, et al., investigated the superheats that could be obtained in liquid hydrogen by rapidly reducing the pressure above the liquid. it appeared that the larger the rate of pressure reduction the larger the superheat that could be obtained, but the rate of pressure reduction depended upon the initial conditions so the effect could have been that of the initial conditions. An acid cleaned glass container was used as an "ideal" system. For a given set of initial conditions, there was no apparent change in -the obtainable superheat when rods of aluminum, stainless steel or brass, with surface roughnesses of from 4 to 78 4in:, were placed in the liquid. A very rough (200 pin, ) stainless steel rod reduced the nucleation superheat. Contaminants, consisting, of solid nitrogen and water particles, also reduced the superheat at nucleation. The theories discussed previously do not correctly predict the results obtained.

15 A number of authors have compared existing hydrogen nucleate boiling heat transfer data with theories and correlations. e.g.,2527,29,35,536.2 A comprehensive work is that of Drayer35 in which 11 equations from the literature are compared with experimental data. The equations predict heat transfer rates, at a given AT, that cover 7-1/2 orders of magnitude. Of those tested, three equations appear to be applicable over a limited range, those of Forster-Zuber,37 ForsterGreif,38 and Cryder-Gilliland.39

II. EXPERIMENTAL APPARATUS A schematic of the experimental apparatus is shown in Fig. 1. The system was designed specifically for the investigation of heat transfer from a solid surface to a saturated pool of a cryogenic liquid. An instrumented test surface was mounted in a holder and immersed into the test liquid which was contained in a vacuum insulated pyrex dewar. Heat was supplied to the test surface by an electrical dc resistance wire heater. Constant vapor pressure over the test liquid was maintained by a pressure control system which vented the vapor as it was formed. A. DEWARS The test fluid was contained in the inner one of the two concentric pyrex dewars shown in Fig. 2. This dewar was designed to be suspended from above by a metal flange attached to a single walled section of the dewar. This section supported the portion of the dewar which held the t-est liquid and which was surrounded by the outer dewar. The liquid section was double walled and vacuum insulated. Its volume was 5 liters with an I.D. and length of approximately 10 cm and 65 cm respectively. The interior of the vacuum space was silvered except for two 1 in. wide strips that were diametrically opposite. The outer dewar was supported from the bottom. It surrounded the liquid portion of the inner dewar and was of similar construction. Liquid nitrogen was placed in the outer dewar to form a heat shield for the test liquid. There was no direct contact between the two dewars. 16

HEATER LEADS ---- SUPPORT TUBE m — PRESSURE RELIEF DEVICE DUMP TUBE P)'ft~Lhf~i~~ —rti~t ~ (CYSOLENOIDVALVE MERCURY SWITC I IP DEWAR RADIATION H () SHIELD - M1 T 1 I — PRESSURE I C LIQUIQ GARGE LSURU _ J 1 WELL MANOMETER RELAY 1110v I Fig. 1. Schematic of test apparatus.

18 When the two sets of unsilvered strips were aligned with each other, it was possible to visually observe the interior of the inner dewar. Reflecting paper was mounted behind the dewars to permit indirect lighting for visual observations. When the unsilvered strips were not aligned, the silvered surfaces formed an effective radiation barrier. All permanent connections (vent line, gas pressurization lines, etc.) to the test dewar were made through a manifold ring mounted on the metal flanges This is shown in Figs. 2 and 3. The manifold ring did not restrict the access to the interior of the test dewar as it had an I.D. approximately the same size as the dewar I.D. This design permitted the test surface to be changed with a minimum of effort as only the connections to the surface itself had to be broken. A dump tube extended through the manifold ring to the bottom of the test dewar. This provided a means of removing the test liquid after a test. By pressurizing the test dewar, the liquid could be forced out through the dump tube and then through a heat exchanger to vaporize it. The vapor was exhausted into the atmosphere. B. PRESSURE CONTROL The controlling device in the pressure control system was a mercury switch. This was a small U-tube, mercury manometer with an electrical contact in each leg. One contact was located at the bottom of its leg so it was always wetted by the mercury and the other contact was located above the normal mercury level. External to the manometer, a relay circuit was wired between the contacts. When the mercury wetted both contacts, the circuit was completed and the relay closed.

19 I_~~~ ~Manifold Ring Single Walled. I r -; Dear >~~u m p~Dm Tube — Outer Dewar Fig. 2. Liquid dewars. Vacuurrm. Line 4 Support' Tube Thermocouple Leads -- Cover Plate Ma.n ifold Ring Fig.:5. Manifold ring and cover plate.

20 The test liquid vapor pressure (dewar pressure) was applied to the manometer leg with the wetted contact and a reference pressure was applied to the other leg. As heat was transferred to the saturated test liquid, its temperature and vapor pressure increased. The pressure increase forced the mercury into the leg with the unwetted contact. When the mercury level reached the contact, the relay circuit was completed and the relay closed, opening a solenoid valve in a vent line. Two needle valves were used in conjunction with the solenoid valve; one in parallel used as a variable vent and one in series used as a flow restriction. The variable vent was adjusted to vent the vapor almost as fast as it was formed. This permitted operation over a wide range of venting rates as the solenoid valve vented only the excess vapor. Tilting the mercury switch from a vertical plane towards the horizontal increased the sensitivity of the pressure control. The actual system controlled the pressure to +0.02 psi. Sparking between the mercury and the make and break contact was at first a problem. A layer of silicone oil was placed over the mercury to prevent the spark from reaching any hydrogen gas that might be present, but the sparking decomposed the oil, contaminating the mercury and resulting in poor wetting of the contacts. A variable resistance was placed in parallel with the mercury between the two contacts and its resistance adjusted so it did not operate the relay, but did offer a lower resistance than the spark gap. This reduced the sparking to a tolerable level. The reference pressure was supplied by a constant mass of nitrogen gas contained in an insulated cylinder. Changes in the ambient temperature produced

21 a slow drift in the reference pressure which, for long runs, was as large as ~0.1 psi. As the liquid was maintained in a saturated condition, this pressure drift is reflected only in small changes in the saturated physical properties of the liquid and vapor. Not shown in Fig. 1 are provisions for purging the reference cylinder and pressurizing it with nitrogen gas. C. SURFACE MOUNTI:NG The original and final designs of the heat transfer surface and its holder are shown in Fig. 4. The difficulties encountered with the original design and the details of the surface preparation are discussed in Section IV.A. In the final design, the heat transfer surface is back by a disc of copper 1/2 in. thick and 1 in. in diameter, which served as a heater block~ A 1 mil thick stainless steel fin was attached to the copper disc to eliminate the physical discontinuity at the edge of the heater surface. A typical stainless steel surface, backed by the copper disc, is shown in Fig. 5. The fin was backed with a Teflon ring to make it more rigid for mounting. Teflon was chosen for the backing material because it has a small thermal conductivity and thus contributes very little to the heat loss. The surface was mounted by clamping the Teflon backed fin and a loading ring between a housing cup and a clamping ring (see Fig. 6-8). A seal between mating members permitted a vacuum to be drawn inside the housing cup. This insulated the back side of the Teflon ring, the heater block and the leads to the surface thermocouple. The members of the clamping assembly were covered with a light coat of vacuum grease to aid in sealing.

22 HEAT TRANSFER SURFACE SOLDER FILLET STAINLESS STEEL FIN LOADING RING CLAMPING RING - - ~ TR- - TEFLON BACKING RING SOLID DISC OF SURFACE MATERIAL COPPER DISC TEFLON HOUSING HEATER HOUSING CUP INSULATION DISC SHEATHED SURFACE -. BACKING DISC THERMOCOUPLE ORIGINAL DESIGN FINAL DESIGN Fig. 4. Schematic of test surfaces and mountings.

Heat Transfer Surface Fin Fig. 5. Test surface. Heater Block Teflon Ring Loading Ring Clamping Ring Fig. 6. Clamping arrangement. Hea Housing C Suppert T Surface ThermocoupX~8 Fig. 7. C ousing cup.

24 Radiation Shields Support Tube Fin Heat Transfer Surface Loading Ring....:a~~~ 0Clamping Ring 8a. Mounted surface without convection shield. Leiad Wires i~rice Cylinder V/ tion Shield:i T i,' { Thermocouple 8b. Mounted surface with convection shield.

25 The loading ring was necessary to maintain the seal in liquid hydrogen. Teflon has a larger thermal coefficient of expansion than the other materials and thus the Teflon ring contracted more than the clamping members when cooled from room temperature to the test temperature. The seal was tight in liquid nitrogen, but at about 400K the differential contraction was sufficient to completely relax the elastic deformation in aLl the parts, The joint then opened and allowed liquid hydrogen to flow into the housing cup, The loading ring acted as a stiff spring during clamping and absorbed most of the differential contraction during cool down. The housing cup provided a location for connecting the heater leads to heavy copper leads. Do SUPPORT TUBE The housing cup was fastened to a support tube. Heavy copper leads for the heater passed through the inside of the tube and into the housing cup. The support tube was bent to position the test surface in the desired orientation and also served as a vacuum line to the housing cup, The support tube passed through the cover plate to the outside of the dewar. Eo COVER PLATE The cover plate was mounted on the manifold ring and sealed the top of the dewar. A sealing gland passed the support tube and allowed it to be raised and lowered as desired, Thermocouple leads entered the test dewar through sealing glands mounted in the cover plate~ A port for the insertion of the liquid hydrogen transfer line was also provided~

26 F. HEATER The heater consisted of 7 mil resistance wire coiled between two layers of glass tape to form a disc 1 in. in diameter. The resistance at the test temperatures was approximately 4'ohms. The leads were 30 gage wire 12 cm long, chosen to approximately balance the I2R heat generation and the heat conduction in the wires. Iron and copper wires were used with liquid hydrogen and liquid nitrogen, respectively. The heater was mounted on the back face of the heater block. It was covered by 1/8 in. of transite insulation and a stainless steel backing disc. Screws passing through the backing disc into the heater block were used to hold the heater and insulation firmly against the heater block. G. POWER SUPPLY The power leads were connected to a circuit designed to give a continously variably voltage of from 0.3 to 356 v. The power was supplied by storage batteries. The maximum power was limited by the heater leads and the amount of liquid vaporized during a test run. HE RADIATION SHIELDS When the surface was mounted to heat in the upward orientation, two radiation shields were mounted on the support tube, in the liquid above the heating surface. These were considered to be necessary as in this orientation the surface faced the cover plate which was at room temperature. When the surface was mounted in the other orientations, it faced the 4 silvered surfaces of the two dewars and no radiation shields were usedo

27 I. CONVECTION SHIELD The convection shield shown in Fig. 8b was used during some of the tests. This is a 1/2 in. length of pyrex tubing mounted in the liquid around the circumference of the test surface, If fluid motion produced by bouyancy forces can be eliminated, a "o-g" environment can be simulated in a "l-g" environment' When heating downwards, the buoyancy forces hold the warm liquid against the surface and if gross convection currents are prevented from sweeping across the surface. a "o-g" environment should be simulated, The convection shield was used to shield the surface from the gross convection currents in the liquid. In the downwards orientation, the mechanism of heat transfer in the liquid should approach pure conduction if convection currents are eliminated. The growth of a vapor nucleus should then require a minimum surface superheat, as the temperature gradient in the liquid is a minimum. The convection shield was always used when a surface was heating downwards. When a surface was heating in an upwards direction, the direction of the buoyancy forces, relative to the surface were reversed, In this orientation, the convection shield was used during some of the tests to prevent the gross convection currents in the bulk liquid from sweeping across the surface. The convection shield was never used with a surface heating in the vertical direction because here it was desired not to interrupt the normal convection pattern.

III. INSTRUMENTATION The measurements recorded were thermocouple EMFs, vapor pressure of the test liquid, voltage drop across the heater and voltage drop across a shunt used to determine the heater current. The measuring instruments are shown in Figs. 9 and 10. Readings not recorded include the reference cylinder pressure and the pressure (vacuum) in the housing cup. A. THERMOCOUPLES The surface temperature was indicated relative to the liquid temperature by a calibrated (see Appendix A) sheathed thermocouple. The 36 gage copper-constantan measuring junction was ungrounded from its 0,040 in. O.D. sheath. The sheath was inserted approximately 1/2 in, into a 0.042 in. hole in the copper heater block. Aluminum dust mixed in a grease binder insured good thermal contact between the heater block and the sheath. The sheath passed through the housing cup and into the liquid through a small hole drilled in the cup. The thermocouple hole in the heater block bottomed at a maximum of 0.032 in. from the surface. The precise location of the thermocouple junction is not important, as at the test temperatures the thermal conductivity of the copper is very large. This, and the relatively small heat fluxes used, results in very small temperature gradients in the copper. The early experiments were conducted with three differential thermocouples located in the Liquid thermal boundary layer at various distances 28

29 ~~~~~~~~~~~~~~~ ~~~~~~~~~~~~~...... ~ It Dewar Pre s sure Gage Mercury Switch -"~ Connections for Test Dewar Fig. 9- Control panel. Amplifier Galvanometer Potentiometer Vo 1 tmet e r Fig. 10. Measuring instruments.

30 from the heater surface. They were made from calibrated 2 mil copperconstantan wire and were mounted in a holder which was seated against the heat transfer surface at opposite ends of a diameter. The measuring junctions were in the approximate center of the surface with the leads running parallel to the surfaces The distances from the surface to the junctions were assumed to'be the same as the distances measure from a flat surface to the center of the approximation 4 mil diameter beads. The distances were measured at room temperature with a calibrated microscope eyepiece scale. The holder was removed when it was determined that the initial vapor was being formed where the holder seated against the surface. The differential thermocouples were referenced to a small copper cylinder located in the liquid at the level of the test surface. The junctions were insulated from the cylinder by finger nail polish on the junctions and a single layer of 7 mil thick glass cloth tape covering the cylinder. The junctions were held in place by several layers of tape and a Teflon ring snapped over the tape. The differential contraction between the Teflon and copper produced a large pressure to hold the junctions against the cylinder in the test,liquid. This was evidenced by deep impressions in the tape at the edge of the Teflon ring. The ends of the cylinder were exposed to the li quid. Also mounted on the reference cylinder was a 30 gage copper-constantan thermocouple, made from calibrated wire and referenced to distilled water ice, and the midpoints of loops of copper and constantan thermocouple wre.s The thermocouple wire loops ran outside the dewar to a potentiometer. As

31 the thermocouple lead wires were slightly cold worked each time a surface was inserted into or removed from the test dewar, the parasitic EMFs in the lead wires increased with use, The thermocouple loops were used to determine the approximate size of the parasitic EMFs and determine when new lead wires should be installed, See Section IV.H for a discussion on the parasitic EMFs, The copper thermocouple lead wires passed through the cover plate to a series of all copper knife switches. These were used to select the input into the potentiometer. B. POTENTIOMETER The thermocouple EMFs were read on a Honeywell Model 2768 6 dial microvolt potentiometer. This instrument has a digital readout in steps of 0.01 Tv. The accuracy in the range of interest is ~(0.01% of Reading + 0.02 pv). C. NULL DETECTION The unbalance from the potentiometer was amplified with a Rubicon Model 3550 photoelectric galvanometer and amplifier system. Building vibrations made it necessary to locate this on a platform suspended from the ceiling. Soft springs in the supports helped to isolate the platform and paddles in oil supplied viscous damping. A Honeywell series 3100 spotlight galvanometer was used to indicate the unbalance. For the surface thermocouple the galvanometer deflection was approximately 100 mm/tv. Even with the amplifier suspended from the ceiling, daytime building vibrations resulted in a noise in the galvanometer of about ~3 mm.

32 D. PRESSURE A calibrated 12 in. Heise bourdon tube gage was used to measure the vapor pressure of the test liquid. The 25 psig range was divided into 0.1 psi steps. The calibrated accuracy of the gage over the entire range was ~0.025 ps:i. Barometric pressure was measured with a mercury barometer. A well type mercury manometer was located in parallel with the Heise gage to measure vacuums during purging operations@ E. HEATER VOLTAGE The voltage drop across the heater was measured by a Hewlett Packard Model 3440A digital voltmeter with a model 3444A plug-in unit. For readings under 10 v the accuracy is ~(0o05% of reading + 0.001 v) and for readings over 10 v the accuracy is ~(0.05% of reading + 0.01 v). The voltage measurements were made at the power supply. The voltage drop in the copper leads (16 gage) running to the heater is negligible~ Fo HEATER CURRENT The heater current was determined by measuring the voltage drop across a resistor in series with the heatero The resistor was a Leeds and Northrup 0o01 ohm precision shunt with an accuracy of ~0.044. The voltage was measured with the same circuit used for the thermocouple measurements.

IV. PROCEDURE Most of the tests were conducted in the same manner. A prepared sure face with heater and surface thermocouple in place was clamped in its mourlting. The surface was then cleaned and the thermocouple connections completed. The mounted surface was placed in the test dewar which was then purged. The test liqui.d was transferred into the dewar and equilibrium conditions were established. After a voltage was applied to the heater, sufficient time was aillowed to establish steady state conditions. Visual observations were made and the heater voltage, the heater current and the thermocouple EMFs were measured and recorded. When the surface was Iheating upwards or vertically' the visual observations were to determine if vapor was being formed at the surface9 and if so, the pattern of the boiling. When the surface was Lheating downwards, the observations were to determine if vapor had already been formed at the surface and filled the convection shield. The heater voltage -was then changed and new reading made. A,o SURFACE PREPARATION' The original design for the heat transfer surface (see Fig. 4), was a round disc of the test surface material mounted in a Teflon housing cup. Teflon was used because of its low thermal conductivity and its large coefficient of thermal expansion. As the surface was cooled down, the Teflon contracted more than the metal and sealed around the surface. This allowed a vacuum to be drawn inside the housing cup to reduce the heat losses.'When 33

34 this design was tested, the initial boiling was at the edge of the surface. The vapor appeared to be forming at the discontinuity between the polished metal surface and the Teflon holder. This prompted an effort to cover up the discontinuity. Data obtained later, with surfaces of the final design, show that vapor forms at a lower superheat on a Teflon surface than on a polished metal surface. It is likely that the vapor formed at the surface's edge was from the Teflon holder, where it was being heated by the metal surface. A thin Layer of vacuum grease was used to cover the discontinuity between the surface and the Teflon holder. This eliminated the boiling at the edge when using liquid nitrogen,'but when using liquid hydrogen the grease developed hairline cracks which served as nucleation sites. About a dozen materials were used to cover the discontinuity, without finding one that would prevent boiling from the edge in liquid hydrogen. A major redesign of the test surface and its holder was made to allow a fin to be attached to the edge of the surface in order to eliminate the discontinuity there. The details of these surfaces are discussed below. The first tests of incipient boiling of cryogenic liquids were tests to study the conditions at which the first vapor would form, while increasing the heat, at an artificial site placed in a polished metal surface. Two 90~ conical diamond indentors, one with a nose radius of 0.0007 in. and one with a nose radius of 0.0001 in., were used to make artificial sites in metal surfaces. These artificial sites did not serve as locations for the initial vapor formation from the surface. If vapor did form from them,

35 there were usually so many other active sites on the surface that it could not be observed. The sizes of the artificial sites tested, the surface materi.al and the liquid in which they were tested are tabulated in Taible II (Section VI.E). When the investigation of the initial vapor formation from artificial sites proved to be impractical, the conditions under which the initial vapor formed from a uniform surface were studied. The surfaces investigated are: CSF-'L Polished stainless steel CSF-3 280 grit lapped stainless steel CSF-4 and CSF-4a 600 grit lapped stainless steel CF-2 and CSF-2a Polished copper CF-4 600 grit lapped copper CTF-i TFE Teflon NASA Glass fiber web with an epoxy cement coating The preparation of the heat transfer surface is extremely important in any study of boiling heat transfer. Seemingly unimportant changes in the preparation can produce large changes in the data. For this reason, the preparation of the surface'will'be discussed in detail. The basic components of a surface were a disc of electrolytic tough pitch copper 1 in. in diameter and 1/2 in. thick and a piece of 347 stainless steel foil i mil thick. Threaded screw'holes were provided on the back face of the copper disc to mount the heater and a thermocouple hole was drilled from the side to within 0.032 in. of the front face. The disc served as a heater -block to smooth out spacial temperature variations present at the resistance heater. The front face of the copper disc was soldered to the foil. The soldering was performed on a lapped surface heated by an electric heater to pro

36 vide control of the temperature and a flat surface on which to work. Care was taken to achieve a bond over the entire face of the disc and to remove excess solder. The stainless steel heat transfer surfaces were made by either polishing or lapping the face of the foil. The polishing was performed with the copper disc mounted n aslowly rotating lathe. The rotating surface was polished with a diametrical motion with cheese cloth dampened by a solution of red rouge dissolved in kerosene. Very little material was removed during polishing and the surface remained slightly wavy from the soldering process. The lapping was performed in a figure eight motion on window glass covered with Clover lapping compound. The lapping was continued only until the foil covering the disc was uniformly lapped. As the foil was not lapped completely through at any point, there was no discontinuity on, or at the edge of, the heat transfer surface. When making a copper surface, a small chamfer was placed on the edge of the copper disc before it was soldered to the foil. The foil and solder on the face of the disc were completely removed by lapping. The solder filling the chamfer held the fin in place. The lapped copper was either tested in the lapped condition or polished smooth in the same manner that the stainless steel was polished. The Teflon surface was made'by coating one side of a piece of foil with 1-1/2 mils of TFE Teflon before soldering it to a copper disc. Micrometer measurements of the coated and uncoated foil were used to determine the thickness of the Teflon coat.

37 The epoxy coated glass fiber web that was tested is the interior lining of the Saturn IV B liquid hydrogen tank. It was carefully removed from its foam insulation backing and glued over a copper surface. When lapping the foil it was possible to observe any spots that were not bonded to the copper disc because these spots pulled away from the disc and lapped faster than the rest of the foil. With practice it was possible to consistently make surfaces which had only 1 or 2% of the surface area unattached. Of the surfaces prepared, approximately one out of every ten was acceptable for use. The main reason that stainless steel surfaces were unacceptablea is that the foil at the unbonded locations would lap completely through before the rest of the foil was uniformly lapped. When making a copper surface, the joint between the copper and fin was often broken when the solder was being lapped from the face of the copper. Visual observations of apparently good copper surfaces sometimes revealed pin holes in the soldered joint that could possibly act as nucleation sites. When a copper surface was tested, if the initial vapor consistently formed at or very near the edge of the copper, the data was rejected and a new surface made. B. ASSEMBLY The assembly began by mounting the heater on the back of the heater block. Then the sheath thermocouple was coated with the aluminum and grease mixture and inserted into the heater block. The surface was positioned, sliding the excess length of the thermocouple through a hole in the housing

38 cup and placing the heater leads inside the cup. After the clamping assembly was loosely fastened, the thermocouple sheath was sealed to the housing cup with Woods metal. Final tightening of the assembly forced the thermocouple against the bottom of the hole in the heater block. The surface was cleaned with a tissue wetted with methylene chloride. When desired, the convection shield and/or liquid thermocouples were positioned. Lead wires were soldered to the thermocouple wires and the junctions, together with the reference cylinder thermocouple and thermocouple wire loops, were mounted on the reference cylinder. A final cleaning of the surface was performed with a cotton swab wetted with reagent grade methylene chloride. The surface was placed inside the test dewar and the cover plate sealed. The thermocouple leads were connected to the all copper knife switches leading to the measuring circuit and the copper leads to the heater were connected to the power source. C. PURGING The system was now ready to be purged. One purge cycle consisted of drawing a vacuum of at least 1/2 cm Hg absolute and then pressurizing with the purge gas to 2 psig. If nitrogen was to be the test liquid, 3 purge cycles were run with nitrogen gas. If hydrogen was to be the test liquid, 2 purge cycles were run with nitrogen gas and then 3 with hydrogen gas. One exception is noted. If nitrogen tests were run with more than one surface, the liquid nitrogen was not emptied from the test dewar while changing the the surfaces. The new surface was held in the vapor space above the liquid for 1 hr so the vapor being vented would provide a purge.

39 If liquid nitrogen remained in the shield dewar from a previous set of tests, it was necessary to keep the surface in the upper portion of the dewar while purging. If the surface was placed in the lower portion of the dewar) the liquid nitrogen would cool the surface and a light frost would then form on the surface from the water vapor in the unpurged air. This made the data unreproducible. D. FILLING The test dewar was filled from portable storage dewars. The liquid nitrogen was purchased locally and supplied in 50 liter flask type dewars. It was transferred into the test dewar through the dump line. The liquid hydrogen was purchased from commercial sources and supplied in 150 liter, superinsulated dewars. It was transferred into the test dewar through a vacuum insulated transfer line inserted through the cover plate. The dip tube extended into the double walled portion of the test dewar. A positive pressure of the purging or4 boil off gas was maintained in the test dewar at all times. This insured that the purge would not be contaminated when connections were being made or during filling. The connections for filling were made from the test dewar towards the supply dewar. A flow of the purginggas through the transfer line purged it as the connection was being made at the supply dewar. The liquid nitrogen in the shield dewar was adjusted to the desired level before filling the test dewar. During testing it was desired to have a flow of heat i~nto the test liquid to maintain it at saturation conditions. When

40o testing with liquid hydrogen the outer dewar was filled with liquid nitrogen but when testing with liquid nitrogen it was filled only about 1/3 full. This gave an inward flow of heat even though the liquid nitrogen in the test dewar was warmer than that in the outer dewar. E. ESTABLISHING TEST CONDITIONS After the test dewar was filled, the pressure control was set to maintain a vapor pressure in the test dewar of 2.5 to 3.0 psig. This pressure was held 1 hr before a test run was begun. Tests showed that 10 min was sufficient for heat transfer from the surroundings to heat the test liquid from its saturation temperature at atmospheric pressure to its saturation temperature at the test pressure. Stratification in the test liquid was not a serious problem. The test dewar had a small volume to area ratio and heat was transferred to the test liquid over most of the surface area. In both liquid nitrogen and liquid hydrogen, a travelling thermocouple probe was used to determine the amount of stratification present. In both cases there appeared to be a slight amount of stratification, but the changes in temperature were within the accuracy of the temperature measurement. After equilibrium conditions were established, the system was pressurized to 5 psig (2 to 2~5 psi over pressure) for 1 min, This was done when a surface was in the downward orientation to condense the vapor that had been trapped in the convection shield. This procedure was followed when surface was in another orientation in order to maintain a consistent pro

41 cedure. With a surface in the upward orientation no change in any data could be detected if the system was not pressurized or if it was pressurized to 10 psig for 2 min. After pressurization, 10 min were allowed'before a heat flux was applied to the surface. Originally a small vacuum line was placed on the surface to remove the vapor from the convection shield when a surface was facing downwards. After the vapor was removed from the surface, liquid would fill the vacuum line and this liquid -was slowly vaporized by heat transferred to it from the surroundings. As the valve on the vacuum line was closed, the vapor eventually'backed up onto the surface. At first this was recorded as very inconsistent data for the initial vapor formation point. F. TESTING A test run consisted of a series of steady state heat flux and temperature measurements along with visual observationso The heat flux was the independent variable and controlled'by setting the voltage drop across the heater, The test procedure was to set a voltage drop across the heater and then to allow sufficient time to establish steady state conditions. The heater voltage, shunt voltage, and thermocouple EMFs were then measured and recorded along with comments on the visual observations. Periodically the slowly varying pressure and EMFs from the reference cylinder thermocouple and thermocouple wire loops were also recorded. After all measurements were completed, the heater voltage was changed and the measurements repeated.

The time needed to establish steady state conditions was determined by stepping the heat flux and recording the surface temperature as a function of time. The criterion for steady state was that no change in surface temperature could be detected over a 5 min period. The time needed varied from a few seconds for a metal surface heating upward in liquid hydrogen to 1/2 hr for a metal surface heating downward in liquid nitrogen. The Teflon surface was a special case and is discussed in Section VIoCo An electric light was used to provide indirect lighting for observing the test surface when testing with liquid nitrogen. When testing with liquid hydrogen, especially with a copper surface, it was observed that the light could initiate boiling. Observations were made in liquid hydrogen'by shining a f~las:hlight.i.nto the dewar but not directly on the surface, This provided suffi.cient light at the surface to make observations but did not appear to initiate boiling. The tests were conducted with the unsilvered strips in the outer and inner dewars aligned. The heat flux was generally begun in the natural convection range and increased in steps through the incipient point. When heating downward, the test had to be terminated at the incipient point, as once the first vapor formed, the entire convection shield filled up with vapor. In the other orientations, if boiling heat transfer datawere desired, the heat flux was increased in steps to a, maximum and then decreased in steps until the surface superheat became so small it could not be measured accurately. Natural convection conditions were established on some surfaces when the heat flux was decreased but vigorous bosiling remained on other surfaces until the test was

terminated. The size of the step in heat flux depended on the portion of the heat transfer curve being investigated. Near the incipient point, the step size was approximately 10o of the heat flux level. Several runs were made.:consecutively with the same surface and liquid charge. If the heat flux was increased immediately after a test was terminated, the last active sites while decreasing the heat flux became active, while increasing the heat flux, at a lower heat flux than would have been necessary if the surface had been maintained for a period of time with no vapor formation. The time necessary to fully deactivate all the surface sites varied from 5 min for the 280 grit stainless steel surface to over 2 hr for the Teflon surface. It was found that pressurizing the system to 5 psig also deactivated the surface sites. After a test was terminated, 15 min were allowed before the system was pressurized to 5 psig and a new test begun. After a series of hydrogen tests with a particular surface was completed, the test dewar was pressurized with hydrogen gas and the liquid hydrogen forced through the dump tube. It was vaporized in a heat exchanger and exhausted into the atmosphere. Sufficient time was allowed for the test dewar to warm up to the nitrogen temperature and then it was purged with nitrogen gas. The test surface was then removed from the dewar and a new one installed. When testing with liquid nitrogen, the liquid was not removed from the test dewar when changing surfaces. The entire system was warmed up to room temperature and cleaned with methylene chloride after two surface changes.

G. VIBRATION TESTS A series of tests were conducted to determine if random building vibrations had an affect on the initial vapor point. A polished stainless steel surface with liquid thermocouples was mounted in the downward orientation in liquid nitrogen and a vibrometer was mounted on the dewar stand. The outputs from the vibrometer and a.'liquid thermocouple were placed on adjacent channels of a strip recorder. The heater voltage was stepped from zero to a value that would give a steady state heat flux approximately 20% above that expected to produce nucleation. The slow temperature transient in the liquid and the output from the vibrometer were recorded. When the initial vapor formed, it produced a blimp in the thermocouple trace. This served as a time mark for determining if vibrations might have caused the nucleation. H. REPLACING THERMOCOUPLE LEADS Care was taken to avoid unnecessary cold working of the thermocouple leads, but some cold working of the leads was unavoidable as a test surface was inserted into, or removed from, the test dewar. This cold working resulted in an increase in the size of the maximum parisitic EMF generated in the wire loops. An effort was made to replace all of the thermocouple lead wires when the EMF generated in the copper loop exceeded 0.4 Ev. A few tests were run after an EMF this large was observed in the copper loop, as it was desired to complete a series of tests. When testing in liquid nitrogen, the average, of the absolute value, of the parisitic EMF in the wire loops was 0. 06 ~v for 25 readings on the

copper loop and 0.64 4v for 11 readings on the constantan loop. When testing in liquid hydrogen,, the values were 0.21 pv for 50 readings on the copper loop and 0.84 p4v for 29 readings on the cqnstantan loop. I. SIURFACE MEASUREMENTS After the heat transfer studies were completed, measurements were made on the surfaces. These included photomicrographs and measurements of su.rface roughnessea The roughnesses were measured first on an instrument that produced a trace of the surface profilde and then on an instrument that gave an RMS reading. The RMS values reported in Section V.F are the average of 4 readings taken in 2 perpendicular directions. Both of the roughness instruments used. a conical diamond tracer with a nose radius of 0e00035 in. and a cone angle of 900. J. DATA REDUCTION An IBM 7090 computer was used to reduce the heat transfer data. From the test data, the computer calculated the po wer into the heater, the surface thermocouple superheat, the heat losses, the net heat flux, a temperature correction for the temperature drop from the thermocouple to the surface and the surface superheat. A discussion on the heat losses and temperature correction is given in Appendix B. The calibration data for each surface thermocouple was fitted with a second order equation over a limited liquid hydrogen temperature range and with another one over a limited liquid nitrogen temperature range. These equations were used to calculate the

thermocouple superheat from the differential EMF measurement, using the saturation temperature of the test liquid, corresponding to the test dewar pressure, as a base. The reference cylinder thermocouple was used only to make sure that the liquid was saturated when data was being recorded.

V. RESULTS A. NATURAL CONVECTION Natural convection heat transfer datawere obtained while determining the initial vapor point. Data from the natural convection measurements are shown in Figs. 1113 The solid lines in Fig. 11 are reasonable fits to the temperatures measured in liquid nitrogen with fine wire thermocouples, with a polished stainless steel surface heating downwards. The dashed lines represent calculated temperature distributions based on the measured surface temperatures and heat fluxes, assuming that conduction is the only mode of heat transfer in the liquid, When a surface was heating downwards in liquid nitrogen, all of the steady state temperatures were very stable with no appreciable oscillations. It was stated in Section III.oI that the purpose of heating downwards was to eliminate the fuid motion produced by bouyancy forces and thus to simulate a "o-g" environment in a "l-g" environment. If all of the fluid motion could be eliminated, the only significant mode of heat transfer in the liquid would be conduction. The two sets of curves in Fig. 11 match reasonably well near the vertical axis, indicating that when a surface is heating downwards with the convection shield, the mechanism of heat transfer in the liquid does approach pure conduction in a appreciable region near the surface. The liquid thermocouples were also used with a surface heating upwards with the convection shield in liquid nitrogen. During natural convection, the superheat of the liquid 0o010 in. from the surface randomly varied from

48 - Run 37-Polished Stainless Steel Surface 2.5 LA- Run 46-Polished Stain2.~ less Steel Surface Uncertainty in location of all liquid thermocouples Heater Convection 2.0 x A Liquid Thermocouple 0 1.0 L L tu - 0.5 T = Tsur-(q/A/kf)x 0.0 0.010.020.030.040.050.060 DISTANCE FROM SURFACE (INCHES) Fig. 11. Temperature distribution in liquid nitrogen when heat downwards.

i2 - Limit of Uncertainty 8. 6 - *.4 Convection ~~~~~~~~~~~~~~~~~~~4 ~~~~~~Shield, 0 d O- H —-Heater SYm. Run Surface Orien. Shield 0 91 Pol. Cu. Upwards Yes 2 rg' Q 92 Pol. S.S. Upwards Yes a 0 93 Pol. S.S. Upwards Yes O 94 Pol. S.S. Upwards Yes o10 - 140 Pol. S.S. Upwards No 8 0A A- V * 141 Pol. S.S. Upwards No 8-AA 0 147 600 Grit S.S. Upwards No 6V V V A 6 V A 37 Pol. S.S. Downwards Yes A V 46 Pol. S.S. Downwards Yes V 146 Pol. S.S. Downwards Yes A 147 600 Grit S.S. Downwards Yes 106 2 6 8 107 2 4 6 8 108 2 4 6 8 109 GrPr Fig. 12. Natural convection heat transfer to liquid nitrogen.

2 102 Limits of Uncertainty 8 6 *0 0 0 VV0 Convection ~o4 A 7Shield lA0 5 Heater 0.19.~~ ~ [J~p~aa~ SP -Run Surface Orient. Shield 102 Pol. S.S. Upwards Yes 2o O 108 Pol. S.S. Upwards Yes O 111 Pol. S.S. Upwards No @ 116 Teflon Upwards No )0 122 600 Grit S.S. Upwards No 101 0 125 280 Grit S.S. Upwards No 8 d 151 Pol. Cu Upwards No'O 158 600 Grit Cu Upwards No 6 A 154 Pol. S.S. Vertical No V 165 280 Grit S.S. Vertical No 4 A 169 600 Grit S.S. Vertical No 106 2 4 6 8 107 2 4 6 8 10 2 4 6 8 109 GrPr Fig. 13. Natural convection heat transfer to liquid hydrogen.

51 10 to 35% of the surface superheat. As the readings were being made with a potentiometer, it was impossible to obtain any accurate measurements of the average liquid temperatures or oscillations. Data for natural convection heat transfer are presented in the form of Nu vs. (GrPr) in Figs. 12 and 13 for liquid nitrogen and liquid hydrogen, respectively. When heating upwards or vertically, the surface temperature oscillated during natural convection about ~4% of the surface superheat, with an unsteady period of 10-30 sec. Each temperature reported under such conditions is near the mean of its observed range. When calculating Nu and (GrPr), the physical properties were obtained from44,47,49,52 and evaluated at the arithmetic mean of the surface and bulk fluid temperatures. Except when heating downwards, the characteristic length was taken as 0.9D as recommended by Krieth5 for a horizontal disc. When heating downwards, the depth of the convection shield was taken as the characteristic length, as this is representative of the thermal boundary layer thickness if there is no fluid motion inside the convection shield and complete mixing outside of It. When one-dimensional conduction is the only significant mode of heat transfer, Nu is unity for all values of (GrPr) if the thermal -boundary layer thickness is taken as the characteristic length. The symbols in Fig. 12 represent three sets of data conditions: the diamonds and squares a surface heating upwards with the convection shield, the circles a surface heating upwards without the convection shield and the triangles a surface heating downwards with the convection shield. Although the diamonds and -the squares represent data obtained with the same geo

52 metrical arrangement, the data appear to be different and were treated as two different sets. The uncertainties in the diamonds and the squares overlap, but at the smaller values of (GrPr) the differences between the surface and liquid temperatures are small and the main contribut:ion to the uncertainties is the uncertainty in the thermocouple calibration~ The same surface thermocoup~le and ca:Librat.i.onr curve -was used for all of the'upward data shown in Fig. 12, so all of the -upwards data at the smaller values of (GrPr) should be in error by about the same amount and in the same direction. At the larger values of (GrPr) the main contribution to the uncertainties is the uncertainty in the heat transfer area. This is ~124 for the copper surface and ~4/ for t'he stainless steel surfaces. The same polished stainless steel surface was used both with and. without the convection sh:ield, so'the data for both conditions at the larger values of (GrPr) should again'be in error by about the same-amount and. in the same di.rection. In addition to the above uncertainties, there is the uncertainty due to the oscillations of the surface temperatures during natural convection heat transfer. The diamonds and the squares in Fig. 12 were each correlated by a least squares fit to a strai.ght line> giving Nu = e105 (GrPr)239 (3) and Nu = 0l46 (GrPr)5335 (4) respectively~ The exponents i.n Eqs. (3) and (4) indicates that the fluid motion was most likely lami:.nar d.ur.i~ng Run 91' and txurbulent during Runs 92, 93,

53 and 94. The transition point given by McAdams43 is (GrPr) = 2 x 107. Background vibrations, as witnessed by the "noise" in the galvanometer used to detect the potentiometer unbalance (see Section IV.C), varied considerably according to the time of day. In view of the facts that the dataare in the transition range and that the'background vibrations varied, it is not surprising that there is a possible change in flow regimes. A 1/4 slope line was fitted to the diamonds and a 1/3 slope line was fitted to the squares, giving Nu = 0.79 (GrPr)l/4 (5) and Nu = 0.15 (GrPr)l/3 (6) respectively. The equations recommended by McAdams43 for a flat plate heating upwards are Nu = 0o.54 (GrPr)l/4 (7) when the fluid motion is laminar and Nu = 0.14 (GrPr)l/3 (8) when the fluid motion is turbulent. Equation (5) predicts heat transfer rates 46% larger than predicted by Eq. (7) and Eq. (6) predicts heat transfer rates 7% larger than predicted by Eq. (8). Equations (5) and (8) are shown in Fig. 12.

54 A comparison of the circles and squares in Fig. 12 indicates that the use of the convection shield increases Nu at a given (GrPr), eo there is an increase in the heat transfer rate at a given ATs. This is most likely because the convection shield disrupts the normal fluid flow pattern. When using the convection shield, the fluid cannot flow towards the heat transfer surface along the fin and therefore a large velocity boundary layer does not build up. As the slow moving fluid in the velocity boundary layer can thermally insulate the surface, the use of the convection shield increases the heat transferred at a given ATs. There are no data available for comparison to Run 91 in Fig. 12, but the effect of the convection shield-would seem to be especially large when the fluid motion is laminar, as the data for Run 91 has approximately 46% more heat transfer than is predicted by Eq. (7). Merte and Clark17 used a similar shield while boiling water in an accelerating system. For values of (GrPr) around 1010, they found that the shield increased the heat transfer rate during natural convection about 50%. The Nu at a given (GrPr) varies considerably between the runs in which the surface was heating downwards. It is noted that Runs 37 and 46 were made with the liquid thermocouples in place. The liquid thermocouple wires and the thermocouple holder, which was seated against'the surface, provided conduction paths for the transfer of heat away from the superheated liquid near the surface. This serves to increase the Nu at a given (GrPr) over runs during which the conduction paths are not present. Some of the variation between runs may also be because the convection shield did not seat exactly the same each time. If superheated fluid could leak past the convection shield,

55 the Nu at a given (GrPr) would again be increased. The symbols in Fig. 13 represent three sets of data conditions: the diamonds a surface heating upwards with the convection shield, the circles a surface heating upwards without the convection shield and the triangles a surface heating vertically without the convection shield. The uncertainty in the data points shown in Fig. 13 is due mainly to the uncertainty in the temperature measurements. For the points with the smaller values of (GrPr), the difference between the surface and liquid temperatures is about OolO1~K not much larger than the uncertainty of ~0.080K caused by paristic EMFs in the thermocouple lead wires. The uncertainties in the temperature measurements remain about the same in magnitude but decrease as a percentage of the surface superheat as the ATs increases. This results in a much smaller uncertainty in the data points at the larger values of (GrPr). A more complete discussion of the uncertainties is given in Appendix B, The very large uncertainty, especially at the smaller values of (GrPr), results in the spread of the data shown in Fig. 13. The same correlations shown in Fig. 12 are shown in Fig. 13 for comparison purposes. The correlation given by McAdams43'for turbulent natural convection heat transfer from a vertical plate is Nu = 0.13 (GrPr)l/3. (9) This equation predicts about 7% less heat transfer than Eq. (8), recommended by McAdams for a horizontal plate under the same conditions. The uncertainties in the data prevent precise calculations, but the data in Fig. 12 show only minor difference between the vertical and horizontal surfaces.

56 No datawere taken with a surface heating downwards in liquid hydrogen, although a number of attempts were made. Vapor continuously formed at the end of the dump tube and on machine screws used in mounting the surface. It proved to be impossible to prevent some of these vapor bubbles from becoming trapped in the convection shield. As power was applied to the heater, the trapped bubbles grew until vapor completely filled the convection shield. B. NUCLEATE BOILING Nucleate boiling heat transfer datawere obtained for a number of surfaces heating upwards and vertically in liquid hydrogen and upwards in liquid nitrogen. No boiling data could be obtained when a surface was heating downwards, as once the initial vapor formed the convection shield filled up with vapor. The surface temperatures during nucleate boiling were much steadier than those during natural convection. oscillating about ~0.5% of the surface superheat. Each temperature reported is near the mean of its observed range. 1. Hysteresis The hysteresis in the surface temperature, which was observed by first increasing the heat flux and then decreasing it, depended upon the surface and orientation. Data for typical runs are shown in Figs. 14-17. The data shown in Fig. 14are for the polished stainless steel surface iheating upwards in liquid hydrogen. This almost complete lack of hysteresis is characteristic of all the runs made with this surface heating upwards. Visual observations indicate that, when increasing the heat flux, a few individual sites distributed rather uniformly over the surface become active

4000 2000 - Run 111-Polished Stainless Steel Surface Heating Upwards 1000 O Increasing Heat Flux 8 W0 - V Decreasing Heat Flux 600- v if0l_ \ \ \ "\\,,Heater 4 400 0 V 0 o 0 - 200- V x 0 -J oV 100 0 w 0 80 8 0 - 0 60 V -O _ Initial Vapor Point V 40- v0 V7 0 Last Vapor Point - V 20 8 10 0 8 * 0.2 0.4 0.6 0.8 1.0 4 6 8 TSUR- TSAT [ K] Fig. 14. Hysteresis of polished stainless steel surface heating upwards in liquid hydrogen.

4000 2000 Run 125 —280 Grit Stainless Steel Surface Heating Upwards O Increasing Heat Flux 0 1000 V Decreasing Heat Flux 600- W\ W Heater O _0l_.0V ~- 400 idlU V o 0' 200 V 0 V V V oV 0 100 80 - 60 0 V 0 40 V 0 0 V 0 I0'c50 Sites — 7V 0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 Ts,,- TSAT [OK,] Fig. 15. Hysteresis of 280 grit stainless steel surface heating upwards in liquid hydrogen.

59 4000 2000 0 Run 122-600 Grit Stainless g Steel Surface Heating Upwards 0 Increasing Heat Flux - 1000 v V0 v Decreasing Heat Flux V A 800 V 0 600 6t\\\\~ —go HHeater ~;. 400 V o o..-,.V 0 - 200 V x 0 X..O X V O0 LL V 0 C2 ---— w Initial Vapor Point 1 00V 80 0 60- 0 V o 0 20 V 0 O -V 0 10 C —---- 16 sites 0.2 0.4 0.60.81.0 2 4 6 8 TSUR- TSAT [ K] Fig. 16. Hysteresis of 600 grit stainless- steel surface heati-ng upwards in liquid hydrogen.

6o 4000 2000 Tefon Surface Heating Upwards O Run 116-Increasing Heat Flux 0 1000 V 0 - Run 116-Decreasing Heat Flux 80~~00 "~0 600 Run 120-Heat Flux Stepped from V Zero to Final Value at Time Zero O X II V I~V ~- 400 I== V Time (min) _1'1~~~~~~~~ _,,1/2 Os 1 Heater 5 0e, v 0 15 35 _ 200 0 1 x 0 V X 80 DL V 20 0 V 0 40 V 20 - ~ _ V Initial Vapor Point 0.2 0.4 6081.0 2 4 6 810 V 0 V 0 Fig. 17. Hysteresis of Teflon surface heaInitial Vapor Point 0.2 0.4 0.6 0.8 1.02 Fig. 17. Hysteresis of Teflon surface heating upwards in liquid hydrogen.

61 first. As the heat flux was increased more, some of the individual sites formed small clusters of 2 or 3 sites which eventually grew larger, joined together and formed patches of active boiling. Other portions of the surface were almost void of active sites with only a few individual sites and clusters. As the heat flux continued to be increased, the patches grew until they completely covered the surface, corresponding to point A in Fig. 14. The site density in the patches was not large and did not increase appreciably until after the patches covered the entire surface. When decreasing the heat flux, the site density decreased fairly uniformly over the surface. At very low values of boiling heat fluxes, a steady site would sometimes have one or two intermittent sites adjacent to it. The fact that the boiling pattern was different when increasing the heat flux and when decreasing the heat flux lead to the expectation that the surface temperature, at a given heat flux, would be different in the two cases and that a hysteresis would be observed. The hysteresis in surface temperature observed with the 280 grit stainless steel surface heating upwards in liquid hydrogen is shown in Fig. 15. When increasing the'heat flux, the active sites again formed first as individual sites distributed rather uniformly over the surface and then as small clusters of as many as 4 and 5 active sites. The clusters did not grow and join up to form patches of active sites as with the polished stainless steel surface. The individual sites and clusters increased in number as the heat flux was increased. Near the maximum heat flux used, the entire surface appeared to be covered with active sites. The site density decreased

62 fairly uniformly over the surface as the heat flux was decreased. As the pattern of boiling was somewhat similar when the heat flux was being increased and when it was being decreased, only a small amount of hysteresis was anticipated. Shown in Fig. 16 are data illustrating the hysteresis in surface temperature observed with the 600 grit stainless steel surface heating upwards in liquid hydrogen. The first vapor came from a patch of active sites which formed suddenly and covered approximately 5% of the surface area. The active site density appeared to be very large in the patch and a drop in surface temperature accompanied the formation of the patch. When the heater power was increased by a step change, the surface temperature was observed to rise slightly for 1 or 2 sec and then to drop as the patch spread. After each increase in heater power, a steady state condition was soon reached in which the patch did not spread further and the surface temperature remained constant-, The steady state temperature was approximately constant until the patch of active sites appeared to cover the surface, corresponding to point A in Fig. 16. When decreasing the heat flux, the site density appeared to decrease uniformly over the surface. The heat flux during patchwise boiling appears to be very non-uniform By assuming that the estimate of the fraction of the surface covered by the patch is accurate, that the portion of the surface covered by the patch transfers heat at the rate measured at the same surface temperature when decreasing the heat flux, and that the remaining portion of the surface transfers heat at the rate measured at the same surface temperature during natural con

63 vection, it is possible to calculate the average heat transfer rate during patchwise boiling to within 15% of the measured rate. The data shown in Fig. 16 and the description of the vaior formation pattern from the 600 grit stainless steel surface also describe the vapor formation pattern from the 600 grit copper surface, heating upwards in liquid hydrogen. The polished copper surface, heating upwards in either liquid hydrogen or liquid nitrogen, first developed a few individual active sites and then suddenly developed a small patch of active sites, similar to the 600 grit surfaces. The site density in this patch was not as large as with the 600 grit surfaces and did not increase appreciably until after the patch had covered the entire surface. Again the site density appeared to decrease uniformly over the surface as the heat flux was decreased. There was a relatively small amount of hysteresis in the data from the polished copper surface. When testing the stainless steel surfaces in the vertical position, bubbles rising from an active site were moved back and forth across the surface by convective currents in the liquid. It was common for potential sites in the path of the rising bubbles to be active for l or 2 sec after the bubbles passed them, and then to become inactive again. When a site on the lower portion of the surface produced vapor steadily, sites in a triangle of surface area above the steady site also became active. This produced a hysteresis, which is somewhat similar to that shown in Fig. 16, in the boiling curve of all the surfaces tested in the vertical position. There was not a single large

64 shift to a lower ATs as the initial vapor formed, but rather a series of smaller shifts as sites on the lower portion of the surface became active and influenced the potential sites located above them. This made the increasing portion of the heat transfer curve rather inconsistent. Visual observations indicated that in the vertical orientation, most of the sites that became active at the lower heat fluxes were on the top half of the surface. This' was most likely because one active site on the lower half of the surface influenced many potential sites on the top half. When decreasing the heat flux in the vertical orientation, the density of active sites was more uniform over the surface then when increasing the heat flux, but was still much greater near the top of the surface than near the bottom. The Teflon surface exhibited the hyster-esis in surface temperature shown by the data for Run 116 in Fig. 17. The sites formed individually with no observed clusters of active sites. All of the sites active at the lower boiling heat fluxes seemed to have difficulty in remaining active. After several seconds or minutes of steady vapor formation at a given site, it was often noted that the site became inactive.; The time between successive data points during the increasing heat flux portion of Run 116 averaged about 10 min. The hysteresis shown was noted with each run made with the Teflon surface, even after it had been in the liquid hydrogen for 24 hr. Upon decreasing the heat flux, the sites were still individual sites and the bubbles produced became very small at low heat fluxes. Also shown in Fig. 17 are the data from Run 120. For this run the power input into the heater was stepped from zero, to its final value, at time

65 zero. The surface temperature was then measured at various times, holding the heater power constant. There were a large number of active sites spread uniformly over the surface area. The surface temperature was decreasing very slowly when the Last measurement was made almost 3 hr after the test -began. As shown'by the data in Appendix C, the surface temperature had decreased only 0.040K during the last 47 min of Run 120. It is felt that if Run 120 had been continued, the surface temperature would have eventually approached the value measured at the same decreasing heat flux during Run 116. The behavior of the surface temperature during Run 120 is felt to'be caused by additional sites becoming active with time. This could not be confirmed'by visual observations as it was impossible to accurately compare the site density over the duration of the test. This effect seems to be cumulative and causes the surface temperature to decrease while the heat flux is increased during Run 116. Comparison of boiling heat transfer data from various runs is made only for the data points obtained while decreasing the heat flux as the decreasing heat flux data is more reproducible and consistent than the increasing heat flux data. Both the increasing and decreasing heat flux data obtained during the boiling runs are tabulated, along with comments from the visual observations, in Appendix C, The data points with very small surface superheats are included in Appendix C,'but' are not plotted in the figures showing boiling heat transfer data because of the large percentage uncertainty pointed out in Section VoA. Also tabulated in Appendix C, but not plotted because of ex

66 tremely large uncertainties in the surface temperatures, are the data from Run 89, made with the epoxy coated glass fiber web surface heating upwards in liquid hydrogen. In general, the data from runs made only to determine the initial vapor point are not included in Appendix C. 2. Reproducibility In order to make the visual observations it was necessary to have the unsilvered strips in the inner and outer dewars aligned. The possibility that radiation through these strips affected the heat transfer data was investigated. Data are shown in Fig. 18 which were obtained with the polished stainless steel surface heating upwards in liquid hydrogen. Run 112 was made immediately following Run 111. The only difference between the two runs is that during Run 111 the unsilvered strips were aligned and during Run 112 the outer dewar was rotated 90o. There is no appreciable difference between the data obtained during Runs 111 and 112. All of the data presented in the other figures were taken with the unsilvered strips aligned. Typical of the excellent reproducibility between runs is that shown in Figs. 18 and 19. The two runs shown in Fig. 19 were made 24 hr apart. The Teflon surface was immersed in the liquid during the entire period between the runs. No datawere obtained during Run 121 while increasing the heat flux. The heat flux was stepped to the maximum value for Run 121 at the conclusion of Run 12Q, shown in Fig. 17, and data taking began immediately on the decreasing heat flux curve. The excellent agreement between the two runs indicates that the Teflon surface temperatures were close to their steady state values.

4000 2000 Polished Stainless Steel Surface Heating Upwards 0 800- Heat Flux Decrease With Successive Data Points O Run 111 —Unsilvered Strips Aligned 600 VRun 112 —0Outer Dewar Rotated V to Reduce Radiation o 4200 0 0 V 2000X~~~ > \ \ Heater X O b 100 0 80 7 60 V 0 V 40- V V - -— Last Vapor Point O O V 20 - 0 0o 10 8'' 0.2 0.4 0.6 0.8 1.0 2 4 6 8 TSuR- TSAT [ K] Fig. 18. Effect of rotating outer dewar on heat transfer to liquid hydrogen.

68 4000 2000 Teflon Surface Heating Upwards Heat Flux Decreases With V Successive Data Points o 1000 0 V 800 O Run 116 0 V VRun 121 0 600 x\\\ I I — vHeater -p" 400 V V 17 0 20 0 - 200 V 0 W V 800 62 V 0 4O V 50 sites VO 8. 0.2 0.4 0.6 0O 1.0 2 4 6 8 10 TSUR- TSAT ['K] Fig. 19. Reproducibility of heat transfer to liquid hydrogen from a Teflon surface.

69 3. Effect of the Convection Shield The effect of the convection shield upon the heat transfer curve is shown by the data in Fig. 20, obtained with the polished stainless steel surface heating upwards in liquid hydrogen. It can be seen that the use of the convection s'hield increases the heat transferred at a given ATs, during bot:h nucleate boiling and natural convection. The disruption of the normal liquid flow pattern was given in Section V.A as a possible explanation for the natural convection portion of the curve. This same argument can be extended to the'boiling portion of the curve. Without the convection shield, the liquid flows towards the heat transfer surface along the fin. This relatively cool liquid can slightly suppress the boiling near the surface edge, This results in a non-uniform heat flux and a decrease in the average heat flux at a given ATs. Another possible explanation is also considered for the nucleate boiling portion of the curve. There is a considerable flow both towards the surface and away from it during boiling. When the convection shield is not used, liquid can flow towards the surface from the sides and vapor away from the surface through the area above it. When the convection shield is used, the flows toward. the surface and away from it both must pass through the area above the surface. This increases the velocities and mixing and perhaps also the heat tra:nsfer coefficient. The difference in the final vapor points shown in Fig. 20 is considered normal for two runs made either with or without the convection shield.

7o TO 4000 2000 Polished Stainless Steel Surface Heating Upwards Heat Flux Decreases With Successive Data Points Vo 1000 VO 800 F V Run 108 —With Con- VO vection Shield 600 0 Run 111-Without 0 Convection Shield V ~ f l C Convection Shield' 777\ Heater V 0 0 V - 200 0 x LL 0 100 80 0 V 60- 0 0 40- V V 0 V 20 ) V Last Vapor Point VO 8,0 l,, 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 TsuRs- TsAT [@K] Fig. 20. Effect of convection shield on heat transfer to liquid hydrogen.

71 4. Effect of Roughness The effect of surface roughness on the boiling heat transfer to liqguid hydrogen is shown by the data in Figs. 21-23. The data shown in Fig. 21 were obtained from three horizontal stainless steel surfaces with different surface finishes: polished, lapped with 600 grit lapping compound and lapped with 280 grit lapping compound. It is seen that the lapped surfaces are more efficient boiling surfaces, i.e., require a smaller surface superheat to transfer a given heat flux, than the polished surface. It is interesting to note that, except for very low heat fluxes, the 600 grit surface is more efficient than the rougher 280 grit surface. The surface superheat necessary to transfer a given heat flux does not decrease monotonically as the surface roughness is increased. A possible explanation for this is discussed in Section VI.B. It is also noted from Fig. 21 that the boiling curves, i.e., the loci of the surface superheat vs. heat flux points, of both lapped surfaces bend towards larger surface superheats at the higher heat fluxes, while the boiling curve of the polished surface does not. This is especially true of the boiling curve of the 600 grit surface. The maximum heat fluxes used were limited by the size of the heater leads and the amount of liquid in the test dewar that could be vaporized during a run. The burnout heat flux was not determined for any of the surfaces, but the magnitude of it can be obtained from values found in the literature. For a horizontal flat surface, Class, et al.,27 report a burnout heat flux of around (4000) x 10-3 watts/ cm2. The data of Mulford and Nigon26 show that the burnout heat flux for a

72 4000 2000 Heat Flux Decreases With Successive Data Points v O 0 Run 111-Polished 1000 Stainless Steel 0 800- V O 800 ~ Run 122 —600 Grit o Stainless Steel 600 - Run 125-280 Grit O Stainless Steel V O 400 0 V IO O V O 80 0 - 200 o 8 0 60 V- J - ~~VO 64~0 0 V0 0 V 1 0 450 sites 16 sites 8 0 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 TSUR- TSAT [ K] Fig. 21. Effect of roughness on heat transfer to liquid hydrogen from horizontal stainless steel surfaces.

73 4000 2000 Heat Flux Decreases With Successive Data Points O Run 154-Polished I000 Stainless Steel V 0 800 V Run 169 —600 Grit 0 Stainless Steel O O 600 <> Run 165-280 Grit Stainless Steel O V O 4000 " t vo V Heater a 0 W o 1 200 0 LL 0 0 " I00 0 - 0 80- 0 0 ~~~60 40. 0 0 0 0 V 0 20 V 0 0.V ~ 15 sites IO- -15 site-s -OV 2 sites TSUR- TSAT [ K] Fig. 22. Effect of roughness on heat transfer to liquid hydrogen from vertical stainless steel surfaces,

74 4000 2000 - V V 1000 VO 800- V 0 - -00 V1 0 V M 0 X V 0 60- V V Run 158-600 Grit.Copper 0 V 40 0 Successive Data Points I 00 0 80 V7 O Run 151 —Polished Copper V Run 158 —600 Grit Copper V 0 40 - V 0 0V 0,~ Heater 2020 V 0'"100 sites 0 0 0 I 0 0 -- -30 sites 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 TSUR- TSAT [ K] Fig. 23j. Effect of roughness on heat transfer to liquid hydrogen from horizontal copper surfaces.

75 horizontal cylinder ranged from (1500) x 10-3 to (6500) x 10-3 watts/cm2. The bending of the boiling curves of the lapped surfaces at the higher heat fluxes is attributed to interference between adjacent sites. The lapped surfaces are saturated with surface cavities and the "area of influence'! of an active-site will overlap potential sites adjacent to it. When two active sites have overlapping "areas of influence," the effectiveness of each individual site is reduced. The density of active sites is larger on the lapped surfaces than on the polished surfaces, and many of the sites that become active at the larger heat fluxes most likely are located within the "area of influence" of sites alreatdy ac-tive. Therefore, a relatively larvae increase in surface tLempera-ture il needed -t'o roduce a criven increase in the heat transfer rate. Shown in Fig. 22 are data obtained from three stainless steel surfaces heating vertically in liquid hydrogen. The surface finishes were the same as those used in the horizontal orientation: polished, lapped with 600 grit lapping compound and lapped with 280 grit lapping compound. The polished and 280 grit surfaces used in the vertical orientation were the same ones used in the horizontal orientation. The 600 grit surface used in the horizontal orientation was damaged before it could be tested vertically, so a new 600 grit stainless steel surface was made and used only for the vertical tests in liquid hydrogen. Again it is noted that except for the low heat fluxes, the;00 g:r'it surface is a more efficient boiling surface than the rougher 280 grrit- surface. The sllape of the boiling curve of the 280 grit surface resembles that of the polished surface more than the boiling curve

76 of the 600 grit surface does. The boiling curves of the lapped surfaces bend slightly towards higher surface temperatures at a much lower heat flux than where the bending took place in the horizontal boiling curves. This might be expected from the observation made earlier that, even at very low heat fluxes, active sites on the bottom portion of the surface have an influence on potential sites located above them. The effective "areas of influence" of the sites on the bottom portion of the surface are very large and overlap even at low heat fluxes. The data in Fig. 23 show the effect of roughness on the heat transfer from horizontal copper surfaces. The 600 grit surface is again more efficient than the polished surface. The bending of the heat transfer curve of the lapped surface to higher temperatures at the higher heat fluxes is especially noticable in Fig. 23. 5. Effect of Orientation The data shown in Figs. 21 and 22 are replotted in Figs. 24-26 to show the effect of orientation on the heat transfer from a stainless steel surface to liquid hydrogen. It is again pointed out that two different 600 grit surfaces were used, one in the horizontal orientation and another one in the vertical orientation. It is seen from Figs. 24-26 that, except for the highest heat fluxes, all of the surfaces have a larger boiling efficiency in the vertical orientation than in the horizontal orientation. This is expected from the observation that when a surface was mounted vertically, potential sites in the path of bubbles rising from an active site became active.

77 4000 2000 Heat Flux Decreases With Successive Data Points 1Q0 Q V Run 1ll —Polished Stainless yo Steel Surface Heating Upwards 800- 0 Run 154 —Polished Stainless Steel V Surface Heating Vertically o 600 V V 200 V X Heater V I00 V 80 V 60 o V 4. 0 0V I00V - 0 O V —- Last Vapor Point V 20 V 0 8 2 sites 0.2 0.4 0.6 0.8 1.0 2 4 6 810 TSUR- TSAT [ K] Fig. 24. Effect of orientation on heat transfer to liquid hydrogen from a polished stainless steel surface.

78 4000 2000 0 V 1000 0 V 8W400F0~ ~ V 800~ 600 0 V 0 400 0 0 - 200 x 0 V IcI00 Ste Surfac Ht l u ar hyiroe frO Heat Flux DecreaseSt W'th Q 00 Successive Data Points 80 O V Run 122 —600 Grit Stainless Steel Surface Heating Upwards 60 V 0 O Run 169-600 Grit Stainless Steel 40 Surface Heating Vertically 0 V 0 20- V"'Heater V 0 V 0 -~15 sites - V 6 sites 0.2 0.4 0.6 0.8 I0 TSuR- TSAT [~K] Fig. 25. Effect of orientation on heat transfer to liquid hydrogen from 600 grit stainless steel surfaces.

79 4000 2000 0 1000 0 V 800V 0 0 v 600 V V0 4 200 D 8C0 V V - Surface Heating Vertically 60 0 200 V'. x 0 Ix _ O V 80O V Run 125-280 Grit Stainless Fig. 26. Effect of orientHeatio on heat trasfer to liquid hydrogen from a 280 grit stainless steel surface. hydrogen from a 280 grit stainless steel surface.

80 At the largest heat fluxes used, the polished surface wasmore efficient at boiling when heating upwards than when heating vertically. It appears that if a slightly larger heat flux was used, the same statement could be made for the 280 grit surface but not for the 600 grit surface. The bending of the boiling curves of the lapped surfaces toward higher surface temperaturesis more noticable in Figs. 25 and 26 than in Figs. 21 and 22. 6. Effect of Surface Material The effect of surface material on the heat transfer from a horizontal surface to liquid hydrogen is shown by the data in Figs. 27 and 28. The same effect is shown in Fig. 29 for heat transfer to liquid nitrogen. In Fig. 27, the boiling heat transfer data obtained with "smooth" surfaces are compared. The Teflon was applied by a plating or spraying process and, although it is smooth, there is no reason to expect it to have the same boiling characteristics as the metal surfaces mechanically polished. Copper is relatively soft and stainless steel is relatively tough, so a consistent manner of surface preparation will produce a different number and size distribution of pits and scratches on copper and stainless steelsurfaces. However, it was anticipated that the boiling characteristics of the two different metal surfaces prepared in the same manner would be somewhat similar. Surface roughness measurements, presented in Section V.E, show that the gross roughness of the two polished surfaces wasabout the same, but photomicrographs, also presented in Section V. E, show more and larger fine surface imperfections on the polished copper surface than on the polished stainless

81 4000 2000 Heat Flux Decreases With Successive Data Points 0 Run 111-Polished V I000 Stainless Steel O O 0 800~ ORun 121-Teflon V O 0 600- V Run 151-Polished Copper V O O 0I400 V 0 400 V o Heater I I X V o0 v 0 X. O 8X O 0 U.3 V 0 60 - 200 V 0 w40pV 0 V XX:~~~~~~ O Last Vapor Point - V 0~ 20 0 - - 0 VO V O 0 10 VO VO 0 50O sites - 30 sites 8 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 TSUR- TSAT [ K] Fig. 27. Effect of surface material on heat transfer to liquid hydrogen from smooth horizontal surfaces.

82 4000 2000 V 0 V 0 1000 0 V 800 0 600 0 V 0 IVR 400 - V O St s 0 0 - 200 x V 0 VJO V 2X ~~0 V~-~~ ~ ~~Heat Flux Decreases With 100 VaSuccessive Data Points 80 V 0 Run 122 —600 Grit 0 Stainless Steel 60 V 40 drogen Run 158 —600 Grit Copper 0 40 V 20 V V4 —100 sites O 0 I0 Q.* —16 sites 8I I - I I A, i J J. I I J I I I I, 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 TSUR- TSAT [OK] Fig. 28. Effect of surface material on heat transfer to liquid hydrogen from rough horizontal surfaces.

83 10,000 8,-000 6,000 Heat Flux Decreases With Successive Data 4000 Points Except for Solid Symbols Which O are for Increasing Heat Flux V Run 91-Polished Copper Heating V Upwards With Convection Shield 2,000 o Run 94 —Polished Stainless Steel V Heating Upwards With Convection Shield. V 1,000 V 800 - V Convection Shield ~ ~600 \ C,~u) |J — Heater V 0 400 9 ____ L Last Vapor Point V — 0 0 807 0 X QLast Vapor Point 400 V I - W 1002I O~~~~~0 80 VV 60 40' 20 0.1 0.2 0.4 0.6 081.0 2 4 6 8 10 Fig. 29. Effect of surface material on heat transfer to liquid nitrogen from smooth horizontal surfaces.

84 steel surface. At a given surface superheat, the heat transfer rate from the polished copper surfacewas as much as 25 times larger then the heat transfer rate from the polished stainless steel surface. Differences in the fine surface finish of the two surfaces may account for part of this difference, but it is difficult to explain the entire difference on this basis. In Fig. 28, data from a 600 grit copper surface and a 600 grit stainless steel surface are compared. Again the surface roughness measurements show that the gross roughness of the two surfaces was about the same. In the lapped condition, as in the polished condition, the copper makes a more efficient boiling surface than the stainless steel. At a given surface superheat, the heat transfer rates from the two surfaces differ by as much as a factor of 20. The data shown in Fig. 29 are for a polished copper surface and a polished stainless steel surface heating upwards with the convection shield in liquid nitrogen. As the copper surface was polished just before each series of tests, the polished copper surfaces tested in liquid hydrogen and liquid nitrogen were not the same surface. In liquid nitrogen, the polished copper surface was again a more efficient boiling surface than the polished stainless steel surface. The difference in surface temperature between the two surfaces was about the same in magnitude when boiling either hydrogen or nitrogen, but wasmuch smaller as a percentage of the surface superheat when boiling nitrogen. 7. Effect of Liquid Shown in Fig. 30 are data from a polished stainless steel surface heat

I0,000.......... 8,000~ 6,000 Polished Stainless Steel Surface Heat Flux Decreases With Succes4,000 sive Data Points Except for Solid Symbols Which are for Increasing Heat Flux 0 2,000 VI Run 108 — Liquid Hydrogen, O With Convection Shield 0 Run 94 —Liquid Nitrogen, With Convection Shield 7 V 1, 000 Convection Shield 0 Heater " V V 0:600v c 400 0, V 0 _O V 0 200 X I V I t. ool v 0 o Last Vapor Point i 0 80 V 60 40- V V. 20 Last Vapor Point V I I 7,,,,,, i, I,,, 0.1 0.2 04 0.6 0.8 1.0 2 4 6 8 10 TSUR- TSAT [OK] Fig. 30. Effect of liquid on heat transfer from a horizontal polished stainless steel surface.

86 ing upwards with the convection shield in both liquid nitrogen and liquid hydrogen. It is noted that that ATs at a given heat flux agree reasonably well for the two liquids in the natural convection region. There is a large difference in the surface superheat at which the last vapor was observed with the two liquids. The slope of the boiling curve is larger with liquid nitrogen than with liquid hydrogen. Data for a 600 grit stainless steel surface heating upwards in both liquid hydrogen and liquid nitrogen are shown in Fig. 31. The boiling curve bends towards larger surface superheats at the larger heat fluxes in both liquids. This occurs at a larger heat flux in liquid nitrogen than in liquid hydrogen. Again the slope of the boiling curve is larger with liquid nitrogen than with liquid hydrogen and there is a large difference in the surface superheat when the last vapor was observed. C. INCIPIENT BOILING The values of heat flux and surface superheat when the initial vapor was observed were measured for 15 combinations of liquids, orientations and surfaces. The results of the measurements from 14 of the combinations are shown in Fig. 32. The one not shown is the surface consisting of a glass fiber web coated with an epoxy cement, heating upwards in liquid hydrogen. One site on this surface could be activated without any heater power by shining a flashlight on the surface. A second site on the surface became active at approximately 1/4~K surface superheat. The symbols in Fig. 32 represent the average of the initial vapor point heat fluxes and surface superheats observed when conducting the tests for the particular point. The lines show

87 laOOO 8,000 O 6,000 600 Grit Stainless Steel 0 Surface Heating Upwards 4,000- Heat Flux Decreases With Successive Data Points Except for Solid O Symbols which are for Increasing Heat Flux 2,000 V Run 122-Liquid Hydrogen O Run 147 —Liquid Nitrogen V V 1,000 V 0 t) 800 Heater"'~i' V -. V 0 rO 400 0 V 0 x 0 XVV 80 V Fig. 31+ Effect last Vapor Point 60V 40- V V 20- 0 V ~25 sites s a 10 III I 0.1 0.2 0.4 0.6 0.8 1.0 2 4 6 8 10 Tsur- Tsar ~'K] Fig. 31. Effect of liquid on heat transfer from a horizontal 600 grit stainless steel surface.

88 4000 No. Surface Orientation Observations 1 Polished S.S. Upward 6 2 Polished S.S. Downward 4 2000- 3 600 Grit S.S. Upward 3 4 600 Grit SS.S Downward 5 5 Polished Cu Upward 1 6 600 Grit S.S. Upward 3 7 Polished S.S. Upward 7 8 Polshed S.S. Vertical 4 1000 9 280 Grit S.S. Upward 3 10 280 Grit S.S. Vertical 3 800 4 11 Teflon Upward 4 12 Polished Cu Upward 4 600- 13* 600 Grit S.S. Vertical 3 14 600 Grit Cu Upward 5 *Different surface than 3, 4, and 6 400 Average q/A and AT OK| >Range of ATS 0 - 200 Individual Observations Fell Along X the Natural Convection Correlation X Shown. The range of Heat Fluxes j ~ ~ Observed is not Shown. I 1O0 / e Nu = 0.14(GrPr) 3 80 ~4// X the: _/O _0 / - Liquid. Liquid /~~~~~ ~ 1Hydrogen Nitrogen 8,,. I.. I _ 0.2 0.4 0.6 08 1.0 2 4 6 8 10 ~.~'TSUR-TSAT ['K] Fig. 32. Initial vapor foration conditions.

89 the range of surface superheats observed. There was a corresponding range of heat fluxes which are not shown. The heat flux and surface superheat of each individual observation was generally located along the natural convection correlations shown. Each symbol in Fig. 32 represents a particular surface and the solid symbols are points obtained with the surface heating vertically. As was pointed out above, the 600 grit stainless steel surface tested in liquid hydrogen in the vertical orientationwasa different surface than used for the other 600 grit stainless steel tests. Also, the polished copper surfaces tested in liquid hydrogen and liquid nitrogenwere different surfaces. There were considerable background vibrations so a series of 5 tests were run to determine if vibrations were influencing the formation of the initial vapor. As described in Section IV.G, the moment of vapor formation on a polished stainless steel surface, heating transiently downwards in liquid nitrogen, was compared with peaks in background vibrations. The average time span between the last vibration of approximately 10% or more above the steady background level and the moment of vapor formation was 29 sec. The smallest time span was 2 sec. This indicates that the background vibrations were not a significant factor in the formation of the initial vapor. When determining the initial vapor point, the heat flux was increased in steps of approximately 10% of its value and steady state conditions established at each level. For some orientations and surfaces, the last conditions without vapor is reported in Fig. 32 as the initial vapor point and

90 for other orientation and surfaces the first conditions with vapor is reported as the initial vapor point. The differences and the reasons will be discussed one at a time. When testing with a surface heating downwards, a relatively large amount of superheated liquid was present on the surface. Once the initial vapor formed, the superheated liquid flashed into vapor and filled the convection shield. In liquid nitrogen, the surface temperature was observed to drop slightly as the superheated liquid flashed into vapor, and then to increase. For points 2 and 4 in Fig. 32, it was thus necessary to report the last conditions observed with no vapor present as the initial vapor point. When heating downwards in liquid nitrogen, the superheated liquid was very stable. It was possible to maintain the liquid superheat at about 75% of the initial vapor point superheat for several hours without forming vapor. No data were obtained heating downwards in liquid hydrogen. As was discussed in Section V.A, vapor was formed in the bulk liquid away from the surface and some of this vapor became trapped in the convection shield, preventing data from being obtained. When conducting the tests included in points 1, 3, 7, and 11 in Fig. 32, the initial vapor was observed to form at individual sites. Often a site would produce a steady stream of vapor bubbles for 3 or 4 sec and then become inactive for several minutes. In an effort to be as consistent as possible with the downwards data, this is reported as the initial vapor point. On the polished surfaces the first vapor was sometimes in the form of individual bubbles that grew very slowly for 2 or 3 sec. The bubbles assumed

91 an almost perfect spherical shape and remained on the surface for 5-10 sec before departing. After a bubble left the surface, a new bubble would form at the same site several seconds later. When conducting the tests for the remaining points in Fig. 32, there was a tendency for active sites to form in small clusters rather than as individual sites. The formation of these clusters of active sites was often accompanied by a drop in the surface temperature, such as shown in Fig. 16. For these points, the highest surface temperature observed without vapor formation is reported in Fig. 32 as the surface temperature when the initial vapor formed. The initial vapor on any particular surface formed at a different site almost every test, indicating that the surface area of each surface was reasonably uniform as far as the initial vapor formation is concerned. In one test with a polished copper surface heating upwards in liquid nitrogen, a felt tipped pen was used to mark on the surface. Numerous sites'became active along the ink marks at a value of surface temperature and heat flux lower than that necessary to form vapor from the polished surface. The data for points 2, 4, and 5 in Fig. 32 were taken with the convection shieild and the data for points 1 and 7 were taken partly with it and partly without it, Comparing the initial vapor points obtained for a particular surface with and without the convection shield, no significant difference could be found. The data for the remaining points in Fig. 32 were taken without the convection shield.

92 The observations composing points 1 and- 7 were made over a period of time while the observations composing the rest of the points were made in a series of consecutive tests. Four series of tests ran over a one week period are included in point 1. The surface was warmed to room temperature and cleaned between each series. The observations for point 7 were made in two series of tests separated in time by three months. The two observations in the second set of tests were both in the middle of the range of the surface superheats and heat fluxes observed during the first set of tests. This indicates that the nucleation characteristics, with respect to the initial vapor formation, of the polished stainless steel surface did not change with time and usage. Comparing points 7 and 9 in Fig. 32 with points 8 and 10, respectively, it is seen that the surface superheat and heat flux, when the initial vapor was observed on a given surface, were about the same when the surface was heating either upwards or vertically in liquid hydrogen. Point 6 and 13 are not directly comparable as two different surfaces were used to obtain them, As shown in Fig. 16, the surface superheat represented by point 6 in Fig. 32 is considerably larger than the surface superheat necessary to form addition active sites on the 600 grit stainless steel surface. If only one site is made when preparing the surface that becomes active at a lower surface superheat, the entire surface will be covered with active sites before the superheat of point 6 is reached. For this reason it is felt that the data for point 13 is more characteristic of what can be expected from a 6o00 grit stainless steel surface than those for point 6.

93 A comparison of points 1 and 3 with points 2 and 4, respectively, provides a comparison of the heat flux and-surface superheat at the reported initial vapor point for a given surface heating upwards and downwards in liquid nitrogen. The heat flux is an order of magnitude less in the downward position but the surface superheat is almost the same. It is again pointed out that when conducting the tests heating upwards, the first point at which vapor was observed is reported as the initial vapor point and that sites were sometimes observed to be active for a few seconds and then to be inactive for minutes. When conducting the tests heating downwards, the last point without vapor is reported as the initial vapor point. Once a vapor bubbleformed when heating downwards, the superheated liquid near the surface flashed into vapor and filled the convection shield. The vapor was held on the surface and could. be observed at any later time. The actual surface superheats when the first vapor formed while a surface was heating downwards and upwardswere closer than those reported in Fig. 32. The initial vapor formation on a given surface heating in liquid nitrogen or liquid hydrogen is primarily a function of surface superheat and is not a strong function of orientation or heat flux' For a given surface heating in a given orientation, the surface superheat and heat flux when the initial vapor forms are reproducible to within +25% of their average values. If point 13 in Fig. 32 is taken as being characteristic of the 600 grit stainless steel surface, except for the Teflon surface, the order of the surfaces determined by the order of increasing surface superheats at the initial vapor point is the same as the order of the surfaces determined by the order

94 of increasing surface superheats at a given boiling heat flux in the upwards orientation. To make the correspondence complete, point 11 in Fig. 32 should coincide with point 10. There appears to be a direct relationship between the ease of forming. the initial vapor from a surface and the boiling efficiency of the surface. D. THEORETICAL ANALYSIS A theoretical analysis for the growth of a vapor nucleus in a saturated pool of liquid will be developed following the analysis of Bergles and Rohsenow.11 As shown in Fig. 33, a vapor nucleus is assumed to completely fill a surface cavity and to extend into the liquid as a hemisphere with a radius equal to that of the surface cavity. The pressure differential between the vapor and the liquid is given by Gibbs equation for the static equilibrium of a vapor bubble Pv Pl- - * (la) rc The Claperon relation in finite difference form, Ap hfg h(2a) AT TsatVfg Tsat relates the slope of the vapor pressure curve to the physical properties. Equations (la) and (2a) are combined to give AT = sat = Tvap Tsat (10) rchfgPg sat

95. y PL Fg. 55. MLIQUID g VAP OR Fig. 33. Model for analysis of the groth. of a vapor nucleus.

96 where Tvap is the saturation temperature of the vapor at Pv and Tsat is the saturation temperature of the vapor at Pi. It is assumed that the temperature in the liquid., near the surface, is represented by T = Tsur - Y (11) kf and that the vapor nucleus grows whenever the liquid temperature at y = rc is equal to or greater than the saturation temperature of the vapor inside the nucleus, obtained from Eq. (10). Combing Eqs. (10) and (11), the range of active cavity sizes at a given surface superheat is (Tsur-Tsat)~(Tsur-Tsat) -4 (hfgpg u-T rmax,min (12) 2 qT kf Using property values of liquid hydrogen at 21~K from44'47'48 and of liquid nitrogen at 80K from,44,49,50 the value of 2aTsat/hfgpg is 1.30 x 10-5 ~K cm and 1.59 x 10-4 ~K cm for hydrogen and nitrogen, respectively. The maximum value of qjkf when the initial vapor formed on a surface is approximately 70'K cm and 2600K cm for hydrogen and nitrogen, respectively. At the initial vapor point 4(q'Ykf)(2oTsat/hfgpg)(l/Tsur-Tsat)2 will always be much less than unity. The square root in Eq. (12) can then be accurately represented by the first two terms of its series representation, giving Tsur-T sat rmax k( 13 ) and

97 rin = 2cTs-t (1-ur (14) The radius of the largest cavity theoretically active at a given (Tsur-Tsat) is the thickness of the thermal boundary layer, defined by conduction in the liquid. For a given orientation, this is approximately constant and independent of the heat flux level. Changes in rmax cannot account for additional sites becoming active as the surface superheat is increased. The radius of the smallest cavity theoretically active at a given (Tsur-Tsat) is independent of the heat flux, i.e., the temperature distribution in the liquid. Because of the physical properties of liquid hydrogen and liquid nitrogen, the vapor nucleus that is theoretically just on the verge of growing, at the initial vapor point conditions observed, is so small that there is no appreciable temperature drop in the liquid between the surface and the outer edge of the nucleus. Thus the vapor growth theories discussed in Section I.B all give the same result for the size of the marginally active cavity for the same assumed equilibrium shape. If the vapor nucleus is assumed to extend into the liquid in some form other than a hemisphere, the radius of the lquid-vapor interface can be geometrically related to the cavity radius. The nucleus will theoretically grow when the vapor-liquid interface has a radius of curvature within the range limited by Eqs. (13) and (14). The cavity size that will theoretically become active at the average initial vapor point conditions and at the knee of the heat transfer curve,

98 observed when heating upwards, was calculated for the various surfaces. The knee of the heat transfer curve is defined as the intersection of the extensions of the fully developed boiling curve- and the natural convection curve. The results of these calculations are presented in Table I. From Table I it is noted that the size of the surface cavities that are theoretically marginally active is extremely small. There is fair agreement between the size calculated for a given surface from the knee data and the initial vapor point data in the same liquid. There is little agreement between the sizes calculated for a given surface from the nitrogen data and the hydrogen data. In every case, the sizes calculated from the hydrogen - data are smaller than the sizes calculated from the nitrogen data. Experimentally, the surface superheat at the initial vapor pointwas almost independent of orientation. The vapor growth analysis predicts that the growth of a vapor nucleus at the initial vapor conditions will be a function of only the surface superheat. For liquids with considerably different physical properties, the analysis depends upon the orientation. Extreme care should be used in extending the results of the initial vapor tests to liquids that have considerably different physical properties than those here. E. ARTIFICIAL SITES Surface cavities were placed in polished copper and stainless steel surfaces with two different 900 conical diamond indentors. The one indentor had a nose radius of 0.0007 in. and the other one a nose radius of 0.0001 in. The potential sites formed with these indenters were not the first active

TABLE I THEORETICAL MARGINALLY ACTIVE CAVITIES Surface Liquid Orientation (q/AxlO3)IVP ( )IVP (dactXlO6)IVp (q/AxlO3)knee (ATs)knee (dactXlO6)knee (watts/cm2) (OK) (in.) (watts/cm2) (OK) (in.) Nitrogen Upwards 347 6.04 20.7 400 6.6o 19.0 Polished Nitrogen Downwards 35.6 4.22 29.6 Stainless Steel Hydrogen Upwards 55.7 1.35 7.8 62.0 1.62 Hydrogen Downwards 48.7 1.39 7.4 Nitrogen Upwards 269 4.59 27.3 200 4.10 30.5 600 Grit Nitrogen Downwards 28.5 3.80 33.0 Stainless Steel Hydrogen Upwards 82.9 2.01 5.1 1.43.53 19.5 Hydrogen Vertical 24.6.74 13.8 280 Grit Hydrogen Upwards 29.2.92 11.1 17.0.66 13.6 Stainless Steel Hydrogen Vertical 23.2.86 11.9 Polished Nitrogen Upwards 295 5.02 24.9 305 4.90 25.6 Copper Hydrogen Upwards 13.2.56 18.3 600 Grit Copper Hydrogen Upwards 17.4.52 20.5 Teflon Hydrogen Upwards 17.4.63 16.2 30.0.92 9.7 ~~~~~~~~~~~~~~~~~~~~~~~~~.9 _ 9.7_._.

100 sites on the surface. If they became active at all, there were so many other active sites on the surface that it was impossible to observe when vapor was first produced at them. Two of the sites appeared to be active while decreasing the heat flux, but it is possible that the cavities were placed on top of naturally active sites already on the surface. The vapor formed at the natural site would then appear to be formed at the cavity. The sizes of the artificial sites tested, the liquid in which they were tested and the results are tabulated in Table II. TABLE II ARTIFICIAL SITES Indentor Cavity Surface Liquid Nose Radius Diameter Results Material (in.) (in.) Stainless Steel Hydrogen.0007.0013 Inactive Stainless Steel Hydrogen.0007.0041 Active when decreasing heat flux to 0.25~K Stainless Steel Hydrogen.0001.00oo46 Active when decreasing heat flux to < O.15~K Stainless Steel Nitrogen.0001.oo0046 Inactive Copper Nitrogen.0001.0042 Inactive Copper Nitrogen.0001.0058 Inactive Copper Nitrogen.0001.0064 Inactive Copper Nitrogen.0001.00oo69 Inactive Copper Nitrogen.0001.0110 Inactive Copper Nitrogen.0001.0185 Inactive F. SURFACE MEASUREMENTS Photomicrographs were made of the various surfaces to obtain information about the microscopic surface finishes. The photomicrographs are shown in Figs. 34-39.

101 500 tin. Fig. 34. Photomicrograph of polished copper surface.. 500 fin. Fig. 35. Photomicrograph of polished stainless steel surface.

102 imii M- l Fig. 36. Photomicrograph of 600 grit copper surface. =..:.. 500 [in. Fig. 37. Photomicrograph of 600 grit stainless steel surface.

103 5}00 fin. Fig. 38. Photomicrograph of 280 grit stainless steel surface. Bare Metal 1000 -in. Fig. 39. Photomicrograph of Teflon surface.

104 It was noted in Section V.B that the photomicrograph of the polished copper surface, Fig. 34, shows more scratches than the photomicrograph of the polished stainless steel surface, Fig. 35. The largest scratches on the copper surface are about 45 pin. wide and the smallest ones visible are 3-4 ptin. wide. Several pits that are from 40 to 200 ftin. in diameter are also visible on the copper surface. The stainless steel surface has scratches in the same size range as observed on the copper surface, but fewer in number and a larger percentage are of the smaller sizes. There are numerous pits on the polished stainless steel surface, ranging in diameter up to 400 kin. Surface imperfections of the size calculated to be active in Section VoD are present on the polished surfaces. There are also larger cavities, that should theoretically be active at a smaller surface superheat. When viewing the lapped surfaces with a stereomicroscope, it was determined-that they were saturated with overlapping surface cavities of approximately uniform size. From the photomicrographs of the lapped surfaces, Figs. 36-38, it is difficult to make any quantitative statements about the size or distribution of the surface cavities because the overlapping cavities are not readily distinguishable in the two-dimensional photomicrographs. Qualitatively it can be said that the 600 grit copper surface, Fig. 36, and the 600 grit stainless steel surface, Fig. 37, have about the same surface roughness. The 280 grit stainless steel surface, Fig. 38, has larger but fewer surface cavities than the 600 grit surfaces. Some portions of the photomicrographs are out of focus because the surface roughness exceeded the depth of focus of the optical system.

105 The photomicrograph of the Teflon surface, Fig. 39, was made at a low magnification because the lack of contrast on the surface did not permit any detail to be observed at a higher magnification. A spot where the Teflon does not cover the base metal is noted. Examination of the surface under the microscope revealed 3 or 4 such spots. Table III contains the results of the surface roughness measurements. Listed for each surface are the average of 4 RMS measurements made with a Profilometer and the average of 15 consecutive peak-to-peak distances measured from a trace of the surface profile made with a Proficorder. Both the Profilometer and the Proficorder were manufactured by Micrometrical Division of Bendix Corporation. The surface traces showed small ripples within the surface cavities, but the tracing instrument was limited by the diamond tracer described in Section IV.I and the readout. The 600 grit surfaces were only slightly rougher than the polished surfaces. The 280 grit surface was considerably rougher than the other surfaces tested. The copper and stainless steel surfaces prepared in identical manners had about the same gross surface roughness.

10o6 TABLE III SURFACE ROUGHNESS RMS Average Peak-to-Peak S (iin.) Distance x 104 (in.) Polished Stainless Steel 5 1.4 600ooGrit Stainless Steel 8 2.4 280 Grit Stainless Steel' 31 31.2 Polished Copper 44 1o4 600 Grit Copper 9 2.3

VI. CONCLUSTONS A. NATURAL CONVECTION From the data shown in Figs. 12 and 13, it is concluded that for a surface heating upwards in either liquid hydrogen or liquid nitrogen without the convection shield, turbulent natural convection heat transfer is correlated well by Eq. (7-8) in McAdams,43 Nu = 0.14 (GrPr)l/3. (8a) The use of the convectyion shield with a small surface increases the natural convection heat transfer slightly when the fluid motion is turbulent and most likely considerably, as demonstrated by Run 91 in Fig. 12, when the fluid motion is laminar. The data from Run 91 is correlated by Nu = 0.79 (GrPr)/. (5a) B. NUCLEATE BOILING The visual observations of the boiling while increasing the heat flux showed that on most of the surfaces there were patches of active sites while other areas of the surface were void of active sites. It is felt that a mechanism described by Corty and Foust7 for the development of the patches is reasonably accurate. According to this mechanism, the larger surface sites lose their vapor nucleus, either because it is displaced as a unit by the liquid or because the vapor diffuses into the liquid. To form the initial vapor, the surface temper.ature has to be increased until a smaller site, 107

108 which still contains a vapor nucleus, becomes active. The vapor generated at the active site spills over into the larger adjacent sites and seeds them with a vapor nucleus. The seeded sites then become active and seed additional sites, forming a patch of active sites. The local heat flux in the patch area increases considerably as the patch forms, so in order to maintain a constant average heat flux the surface temperature must drop. The patch will stop growing when there are no potential sites within the seeding range of the active sites that will become active at the surface temperature. When the average heat flux is increased, the surface temperature rises momentarily and then the patch begins to grow again. For the proposed mechanism to work, it is necessary for potential sites to be located close together. The lapped surfaces are saturated with surface cavities and satisfy this requirement while the photomicrographs of the polished surfaces show numerous scratches that could be used to spread a patch. Three surfaces did not develop patch boiling, the Teflon surface, the epoxy coated glass.bter surface and the 280 grit stainless steel surface. The Teflon and epoxy were applied by spraying or plating and most likely have the sites well separated. The 280 grit stainless steel surface will be discussed in detail below. Good and Ferry41 report that liquid hydrogen had a zero contact angle on all the materials that they investigated, including stainless steel and Teflon. The theory of Bankoff,l4 predicts that liquid will replace vapor in all the surface cavities when the liquid completely wets the surface. It is postulated that, due to the extreme wettability of liquid hydrogen, the

109 liquid replaces any vapor that may be introduced into the larger surface cavities. This is especially important with the 280 grit stainless steel surface as this surface contains many cavities that are larger than most of the cavities on the other surfaces investigated. This does not exclude the possibility that secondary sites located within the surface cavities will be active. If the liquid does wet many of the cavities on the 280 grit stainless steel surfacer, it might be expected not to develop patchwise boiling. After one of the smaller surface cavities or a secondary site located within a surface cavity becomes active, vapor will be introduced into adjacent surface cavities. A small cluster of active sites might form, but a lack of surface cavities adjacent to it that will retain a vapor nucleus prevents the formation of a patch. There may be secondary sites inside the inactive surface cavities that will become active at a higher surface temperature. It was observed that on a vertical surface heating in liquid hydrogen, sites in the path of the bubbles rising from a steady site become active for a few seconds after the bubbles passed by them, It is felt that there was a tendency for these intermittent sites to be wetted by the liquid hydrogen and to lose their vapor nucleus. The seeding of these sites by the rising vapor permitted them to be active until the vapor nucleus was lost again. If the measured. surface cavity sizes in Table III are compared with the results in Table I of the site sizes calculated to be marginally active, it can be seen that secondary sites located within the surface cavities may

110 make a major contribution to the number of active sites on all of the surfaces. It was noted in Section V.B that the boiling characteristics of the 280 grit stainless steel surface were consistently more like the characteristics of the polished stainless steel surface, than were those of the 600 grit stainless steel surfaces. This is likely if liquid filled many of the surface cavities on the 280 grit surface. The roughness as far as boiling is concerned would then be the roughness of the inside of the cavities. For boiling, the 280 grit surface may be smoother than the 600 grit surface. Mulford and Nigon26 boiled liquid hydrogen from a copper cylinder and found no change in the boiling characteristics of the cylinder when it was roughened by sand blasting. It is possible that the sand blasting produced surface cavities that were wetted by the liquid hydrogen, and that the cavities serving as active sites were left unchanged. The difficulty of making active artificial sites was noted in Section V. Eo These surface cavities were relatively large because they were made by a mechanical means. If the liquid wetted them, they would not serve as active nucleation sites. This is most likely why the artificial sites were not active. The copper surfaces were much more efficient as boiling surfaces than the stainless steel surfaces prepared in an identical manner. The reason for this not known, bt itis speculated that the liquids wet the copper slightly less than the stainless steel~ This means that some surface cavities on the stainless steel surfaces that are wetted by the liquid, would not be wetted

if they were on a copper surface. They would serve as active sites on the copper surface but are not active sites on the stainless steel surface because they are wetted with the liquid. A given surface is more efficient at transferring heat when heating vertically than when heating upwards. Vapor bubbles rising from an active site tend to seed potential sites with vapor and to make them active. Some of these seeded sites remain active indefinitely and some are active only a few seconds, but in either case the result of additional sites is to increase the heat transferred at a given surface temperature. C. INCIPIENT BOILING The formation of the initial vapor on a surface heating in liquid hydrogen or liquid nitrogen is primarily a function of the surface superheat and is not a strong function of the heat flux or orientation. As the bouyancy forces change direction relative to the surface when the orientation of the surface is changed from upwards to downwards, the bouyancy forces do not appear to be a significant factor in the formation of the initial vapor in hydrogen or nitrogen. The surface superheat and heat flux when the initial vapor was observed on a given surface in a given orientation were reproduci'ble within ~..25% of their average values. There is a direct relationship between the ease of forming the initial vapor on a uniform surface and the boiling efficiency of the surface when heating upwards. In general, the lower the surface superheat at the initial vapor point, the lower the surface superheat needed to transfer a given boiling heat flux.

112 The theoretical analysis of the growth of a vapor nucleus indicates that the nuclei that grow in liquid hydrogen and liquid nitrogen at the observed initial vapor conditions are very small. Because of the physical properties of these two liquids, the growth of vapor nuclei at the initial vapor conditions is theoretically independent of the heat flux and temperature distribution in the liquid, depending only upon the surface superheat. Extreme care should be used in extending to other liquids which have considerably different physical properties..the conclusion that the initial vapor formation in liquid hydrogen and liquid nitrogen is not a strong function of orientation or heat flux. D. RECOMMENDATIONS FOR FUTURE WORK Perhaps the, most promising. area for future work is an investigation to try and determine the optimum surface preparation for obtaining an efficient boiling surface, especially for liquid hydrogen. There is not reason to believe that the 600 grit surfaces represent optimum boiling surfaces for liquid hydrogen. Surfaces lapped with a coarser lapping compound than the280 grit compound used here should also be investigated in liquid hydrogen to see if their boiling characteristics are more like those of the polished surface than those of the 280 grit surface are., The investigation of the effect of surface material upon'boiling and the initial vapor- formation should be continued. Copper surfaces are much more efficient at boiling liquid hydrogen and form the initial vapor at a ulower surface superheat than stainless steel surfaces prepared in-the same manner,

113 The data for the Teflon surface were between the data for the two polished metal surfaces. The data from these three surfaces do not follow any consistent pattern that can be related to the thermal-physical properties of the materials, but this might be because of the physical surface finish of the Teflon surface. More data is needed with. other metal surfaces before a pattern can'be determined.

APPENDIX A THERMOCOUPLE CALIBRATION The accuracy desired in the temperature measurements indicated that the thermocouples should be calibrated. One of the fine wire liquid thermocouples and one of the 30 gage reference junction thermocoupleswere calibrated and considered to represent the calibration for all the thermocouples made with the same wire. Each of the sheathed thermocouples, used to measure the surface superheat, was calibrated separately. A schematic of the calibration apparatus is shown in Fig. A-1. The system is similar to the test system, with the glass test dewar replaced by a stainless steel cryostat. When the cryostat was filled with a cryogenic liquid, a constant pressure was maintained in the cryostat by the pressure control described in Section II.B. The reference pressure cylinder was immersed in an ice bath to maintain a constant temperature, removing the drift in pressure that was reported for the heat transfer tests. A mechanical stirrer was used to prevent stratification in the cryogenic liquid. The thermocouples to be calibrated were imbedded in a copper cylinder located inside the cryostat. The thermocouples were referenced to a distilled water ice bath. The EMFs generated were measured on the potentiometer circuit described. in Sections IIIoB and III.C. The copper cylinder contained a small equilibrium cell which could be charged with a high purity gas. The gas passed through a flow meter before 114

"PRESSURE RELIEF DEVICE PRESSURE GAGE -1Om FLOW METER -SOLENOID STIRRER DRIVE TO POTENTIOMETER SOLENOID VALVE FILL PORT IM11RCURY SWITCH 0 3 S EBATH P/JIGLIQUID LEVELN ]PROBE R l (I Fig — Sceatcofth ulob ~i!~ ISTAINLESS C3 STEEL CRYROSTAT 0 o r> SUPER " I INSULATION t~ LJCLIDRTo VACUUMJ PUMP COPPER a. ~CYLINDER'T x: a:, ICE BATH WELL MANOMETER 00 EQUILIBRIUM. MECHANICAL STIRRER -J; LIQUID LEVEL PROBE RELAY III0v Fig. A-1. Schematic Of thermocouple calibration apparatus.

116 entering the equilibrium cell. The pressure in the equilibrium cell was measured with a calibrated Heise bourdon tube gage if it was above atmospheric pressure, and with a well type mercury manometer if it was below atmospheric pressure. The 12 in. Heise gage had a 0-100 psi range with divisions of 0.1 psi. The calibrated accuracy over the entire range was ~0.1 psi. The temperatures of the calibration points were limited to the temperatures between the triple point of a liquid and its saturation temperature at 100 psig. For this work, two sets of data were taken, one in the liquid nitrogen range and one in the liquid hydrogen range. They were used as separate calibrations when conducting the heat transfer tests. The calibration began by purging the cryostat with gaseous helium and the equilibrium cell with high purity nitrogen or hydrogen gas, depending upon the temperature range of the calibration point desired. The cryostat was filled with a cryogenic liquid which was maintained at a constant pressure. The temperature of the calibration point was equal to the saturation temperature of the cryogenic liquid. When the temperature of the copper cylinder became steady, as shown by the thermocouples, the high purity gas was slowly charged into the equilibrium cell. The equilibrium cell pressure rose steadily until the saturation pressure of the gas was reached. The pressure then held steady while the flow meter showed a continuing flow, indicating that the gas was being condensed. The gas flow was shut off and 10 min allowed to dissipate the energy transferred to the copper from the gas. The equilibrium cell pressure and thermocouple EMFs were then measured and recorded.

117 The temperature of the copper cylinder, and the thermocouple junctions,f was determined from the saturation pressure in the equilibrium cell and the vapor pressure data given by Scott.44 For hydrogen, the normal hydrogen vapor pressure data was used. The calibration data in the nitrogen temperature range was plotted directly as EMF vs. temperature. The sensitivity was the same for all of the thermocouples, approximately 16 1tv/OK. The maximum deviations of the calibration data from the curves drawn were approximately ~0.5 pv or ~t0.03K. The maximum uncertainty in the temperature, due to uncertainty in the pressure reading is ~0.040K. The combined uncertainty is ~0.05~K. The decreased sensitivity of the thermocouples in the liquid hydrogen temperature range, approximately 6 ~v/~K, necessitated more care in the reduction of the calibration data. For each thermocouple, deviations of the measured EMFs from the standard values given by Powell and Bunch4 were calculated. These deviations were plotted against the measured EMFs and a straight line was fitted to the points by a least squares method. This line was assumed to represent the exact deviation and used to construct an EMF vs. temperature curve for the thermocouple. Except for the fine wire thermocouple, all of the deviations were within ~0.5 Pv of the fitted line. The deviations for the fine wire thermocouple were all within ~1.0 -v of the fitted line. For the sheath thermocouple with the largest variations in the data, the mean squared variation from the fitted line was 0.032 (v) 2. A 99' confidence interval on the slope of the line is +.0057,v/iv. This is reflected in the EMF vs. temperature line as ~0.03 pv/~K or about ~0.50,.

APPENDIX B ERROR ANALYSIS 1. HEAT FLUX The average heat fluxes reported contain errors and uncertainties from at least four sources. The sources considered are: 1. Error in the heater current measurement. 2. Error in the heater voltage measurement. 3. Error in the heat transfer area, 4. Error in the calculated heat losses. The heater current was determined by measuring, with a potentiometer, the voltage drop across a precision shunt located in series with the heater. In the range of these measurements, the accuracy of the potentiometer is ~0.01%, but the readings were made only to ~0.1%. The accuracy of the shunt resistance is +0.04%. Combining the uncertainties, the heater current is known to ~0.108%. The heater voltage was measured by a digital voltmeter with an accuracy of:(O. 05% of reading + 0.01% of full scale). The maximum uncertainty in the heater voltage was ~0.25%. The uncertainty in the heat transfer area depends upon the type of surface. The copper disc backing a stainless steel or Teflon surface had square edges on the face that was soldered to the foil. A solder fillet was not required with these surfaces as the foil was soldered over the entire heat transfer area. The fillet remaining after soldering was about O.010 in. 118

119 This gives an uncertainty in the heat transfer area of ~4%. A chamfer of about 0.030 in. was placed on the edge of the copper disc used to make a copper heat transfer surface, The solder fillet was required to hold the fin in place as this was the only physical bond to the fin.e A fillet of about 0~030 in. was used, giving an uncertainty in the heat transfer area of +12*. The total energy input into the heater was reduced by the amount of energy calculated to -be conducted away from the surface'by the heater leads, the thermocouple wires and the fin and its *backing. The temperature of the fin decayed to the liquid temperature in about 2 mm, so the effect of curvature could be neglected. Two dimensional calculations showed that the contribution of the Teflon backing ring to the total heat loss was negligible when the stainless steel fin was between it and the liquid. A layer of solder 0o0001 in. thick was assumed to have remained on the fin when the liquid solder was wiped from it. This solder did make an important contribution to the total heat loss. When calculating the heat losses, the heat transfer coefficient on the fin was assumed to be constant. Values of 35 x 10-3 watts/cm2 ~K and 40 x 10-4 watts/cm2 OK were used for liquid hydrogen and liquid nitrogen, respectively, in both the upwards and vertical orientations. These values are characteristic of the natural convection data. The heat transfer coefficient on the fin, when heating downwards in liquid nitrogen, is difficult to estimate. A value of 20 x l0-3 watts/cm2 ~K was used in the heat loss calculations.

120 The model used to calculate the fin heat loss is shown in Fig. B-1. h,Tsat STAINLESS STEEL El k, SOLDER 2 k Tsur Fig. B-1. Model for fin heat loss calculations. Lunping the temperature across the thickness of the solder and stainless steel, the temperature distribution in the fin is given by 1/2 (T-Tsat) = (Tsur-Tsat )e (B-l) The total heat loss from the fin is (qL)fin = D TD(Tsur-Tsat) [h (k11+k252) ] (-2) The thermocouple wires were assumed to be insulated when they were inside the housing cup and at the temperature of the liquid where they passed into the liquid. The heater leads were also assumed to be insulated and at the temperature of the liquid where they attached to the heavy copper leads. The loss in each wire is then by conduction down its length and can be written in the form (qL)wire = kA(Tsur-Tsat/L) (B-5)

121 The total calculated heat losses are listed with the data in Appendix C. The percentage of the total heat input that they represent depends upon the surface superheat, the orientation and the liquid. For example) the calculated heat loss from the polished stainless steel surface) heating upwards in liqui.d Ih.ydrogen at the initial vapor conditions, is approximately 11% of the total heat input. This percentage increases at smaller heat fluxes and decreases to about 1% during vigorous boiling. The heat loss calculations are considered to be accurate to ~25%. For the case considered above, this represents an uncertainty in the reported heat flux of ~2.8%. The total uncertainty in the heat flux is a combination of the separate uncertainties and depends upon the surface, heat flux, orientation and liquid. Again using the example of the polished stainless steel surface, heating upwards iJn gliquid hydrogen at the initial vapor conditions, the total uncertainty in the heat flux is ~5o1%o 2. TEMPER4ATURE MEASUREMEETS Five possible sources of error in the reported temperatures are considered, They are: Error in the thermocouple calibration. 2, Error in the EMF measurement. 3. Parasitic EMFs in the thermocouple leads. 4. Error in the temperature correction. 5 Cooling of the thermocouple junction by conduction of heat in the thermocouple wires.

122 The thermocouple calibration was discussed in detail in Appendix A. There it is estimated that the accuracy of the calibration is within ~0.5% for the differential thermocouples in liquid hydrogen and +~0.05K for thermocouples in liquid nitrogen, referenced to ice. For a differential thermocouple in liquid nitrogen this would be an uncertainty of ~0.07'K. The thermocouple EMFs were measured with the potentiometer circuit described in Sections III.B and III.C. For the polished stainless steel surface, heating upwards at the initial vapor conditions, the potentiometer accuracy of ~(0.05% of reading + 0.02 Ctv) represents approximately ~0.25% of the surface thermocouple's EMF in liquid hydrogen, and ~0.07% of the surface thermocouple's EMF in liquid nitrogen. The percentage increases at smaller surface superheats and decreases at larger surface superheats. The parasitic EMFs in the copper thermocouple leads add to or subtract from the thermocouple EMF. It is felt that the parasitic EMFs were almost always within the range of +0.5 Tv. This represents +~.080K and ~0.03~K when measuring temperatures in liquid hydrogen and nitrogen, respectively. Except for the nonmetallic surfaces, the calculated correction to the measured temperature, to account for the temperature drop between the thermocouple and the surface, is not large. It is directly proportional to the heat flux and listed with the data in Appendix C. For the polished stainless steel surface, heating upwards at the initial vapor conditions, it represents about 1% of the surface superheat in both liquid hydrogen and liquid nitrogen. Assuming the calculated correction to be accurate within +25%, this is an uncertainty of ~0.23% in the reported surface superheat.

123 An estimate has been made of the temperature difference between the thermocouple junction and the heater block, resulting from conduction in the thermocouple wires. This difference is estimated to be 0.7% of the surface superheat in liquid hydrogen and negligible in liquid nitrogen. The total uncertainty in the surface superheat depends upon the surface, the liquid and the heat flux. For the polished stainless steel surface,heating upwards at the initial vapor conditions, the uncertainty is ~6.1% and ~1.77 in liquid hydrogen and nitrogen, respectively. 3. POSITION OF LIQUID THERMOCOUPLES In order to determine the location of the liquid thermocouples, the thermocouple holder was seated against a flat surface and the distances from the thermocouple beads to the surface were measured at room temperature, with a calibrated microscope eyepiece scale. The thermocouple wires were 2 mils in diameter and the soldered thermocouple beads were approximately 4 mils in diameter. The distance measurements were made to the approximate center of the bead with an accuracy in the reading of ~0.001 in. This is combined with the uncertainty resulting from the relatively large size of the beads to give an uncertainty in the locations of the tremperatures measured of ~0.0022 in. The relatively small contraction of the holder when cooling from room temperature to the test temperatures is negligible.

APPENDIX C DATA The data from valid runs during which boiling heat transfer data were obtained, plus a few runs in which only natural convection datawere obtained, are tabulated below along with comments from the visual observations. The pressure indicated is an average at the test surface over the duration of the run. This pressure changed slightly during a run because of drift in the reference pressure and a changing hydrostatic head as liquidwas vaporized. The average liquid depth is the average amount of liquid above the surface. During a long run, the liquid level dropped as much as 8 in. The measured power input into the heat is and the calculated heat loss is qL. qM is reduced by qL and then divided by the area to obtain the net heat flux, qnet/A. bTorr is the calculated temperature drop from the surface thermocouple to the surface. ATs is corrected by 6Tcorr and is equal to the surface superheat (Tsur-Tsat). 124

125 Run 37 Date 10/18/66 Surface Solid Polished S.S. Liquid N2 Pressure 14.0 psia Orientation Downwards Average Depth of Liquid Convection Shield Yes qM x 103 qL x 103 A Tcorr ATs Comment (watts) (watts) (watts (OK) (OK) cm2 49.0 13.9 6 95 0.01.81 Distance from surface =.000 in..74.007.64.015.55.035 75.6 20.6 10.8 0.01 1.28.000 1.16.007.99.015.84.035 84.0 21.9 12.2 0.02 1.37.000 1.25.007 1.07.015.90.035 92.0 24.3 13.4 0.02 1.50.000 1.38.007 1.17.015.99.035 104 26 15.4 0.02 1.64.000 1.51.007 1.29.015 1.08.035 113.5 29.2 16.8 0.02 1.83.000 i1.67.007 1.41.015 1.19.0355

126 Run 46 Date 10/21/66 Surface Solid Polished S.S. Liquid N2 Pressure 13.9 psia Orientation Downwards Average Depth of Liquid Convection Shield Yes 13 x 3 cnet x 103Comments qsx2, 0S.L 105 A 1 Tcorr ATs Comments (watts) (watts) (watts) (OK) (OK) cmR 144.2 36.3 21.3 0.03 2.27 Distances from surface =.000 in. 2.07.007 1.74.015 1.43.035 160.9 40.1 23.9 0.03 2.51.000 2.27.007 1.90.015 1.55.035

127 Run 89 Date 3/28/67 Surface NASA Liquid N2 Pressure 17.0 psia Orientation Upwards Average Depth of Liquid Convection Shield Yes qnet x 103 qM x 103 qL x 103 A 3Tcorr ATs (watts) (watts) (watts) (OK) (OK)ments cm2 19.0 4.2 3,7 0.11 0.10 From Run 90-1 active site 43.8 9.8 6.7 0.20 0.29 1 active site 65.2 14.4 10.0 0.30 0.42 From Run 90-1 active site 70.8 15.4 10.9 0.33 0.44 2 active sites 20.8 14.9.45 0.59 Some sites with large bubbles and low frequency and some 133 28.4 20.5 0.62 0.80 sites with small bubbles and high frequency 172 36.4 26.8 o.80 1.02 200 42.2 31.1 0.94 1.17 249 52.2 38.9 1.16 1.45 309 64.2 48.2 1.45 1.76 384 79.2 60.2 1.86 2.10 511 104 80.3 2.41 2.79 660 133 104 3.12 3.54 Site density increasing noticably 824 165 130 3.90 4.34 1380 27Q 219 6.57 6.91 2110 399 338 10.1 10.0 2960 538 478 14.4 12.6 4040 694 660 19.8 14.9 5890 940 976 29.4 17.6 4520 759 741 22.3 15.7 3560 627 578 17.3 14.0 2770 506 447 13.4 11.9 1710 325 272 8.16 8.08 1040 199 165 4.97 4.96 591 117 93.6 2.84 2.99 421 84.o 66.5 2.00 2.20 274 55.2 43.2 1.29 1.47 193 39.2 30.4 0.91 1.05 137 28.2 21.5 o.65 o.86 101 21.0 15.8 0.47 o.58 78.1 16.2 12.2 0.36 0o.45 45.4 9.8 7.0 0.21 0.28 Note: qL and 5Tcorr are only guesses as the thermal properties of the glue used were not available. This run was not plotted because of the large uncertainties, but is included here for completeness.

128 Run 91 Date 1/12/67 Surface CF-2a Liquid N2 Pressure 17.1 psia Orientation Upwards Average Depth of Liquid Convection Shield Yes q nex 103 qM X 103 qL x 03 A x 10 Tcorr ATs (watts) (watts) (Watts) (OK) (OK) Comments cm, 42.5 7.36 6.80 0.00 0.24 70.2 11.5 11.5 0.00 0.37 109 17.2 18.1 0.00 0.55 177 25.6 29.9 0.00 0.82 257 34 9 43.8 0.00 1.12 435 54.3 75.1 0.00 1.74 680 77.7 119 0.00 2.49 965 103 170 0.00 3.29 1340 135 239 0.00 4.31 1650 157 295 0.00 5.02 2 steady sites right together 158~0 151 28~ 0.00 4.84 1 steady site-bubble size Much smaller than above 1530 146 273 0.00 4.68 Site intermittently active 1470 142 262 0.00 4.55 1390 138 247 0.00 4.41 1360 135 242 0.00 4.34 Site inactive 1390 138 247 0.00 4.43 1470 143 262 0.00 4.57 Same site active again 1980 175 356 0.01 5.59 1 group of 3 sites and 1 of 4 2280 168 417 0.01 5.36 A small patch, not at above sites 2990 185 552 0.01 5.92 4070 179 767 0.01 5.71 5560 190 1060 0.02 6.o6 8130 200 1560 0.03 6.37 11000 190go 2120 0.03 6.o5 14100 217 2730 o.o4 6.90 16100 216 3130 0.05 6.87 19800 225 3860 o.o6 7.14 17200 217 3350 0.05 6.89 14500 208 2830 0.05 6.63 11600 200 240 o.o04 6.36 9480 193 1830 0.03 6.16 7960 185 1530 0.02 5.90 65oo 184 1250 0.02 5.89 5330 179 1020 0.02 5.73 3870 170 729 0.01 5.43 2900 159 541 0.01 5.07 1810 128 332 0.01 4.10 1210 108 217 0.00 3.47 2 sites 853 89.5 151 0.00 2.87 no sites 362 47.1 62.1 0.00 1.51 89.o 15.6 14,5 0.00 oo 0.50

129 Run 94 Date 4/25/67 Surface CSF-1 Liquid N2 Pressure 17.2 psia Orientation Upwards Average Depth of Liquid Convection Shield Yes 3net x 103 qM x 103 qL x 103 A 3Tcorr ATs (watts) (watts) (watts) (~K) Comments cm 89.0 17.1 14.2 0.00 0.55 Taken from Run 92 137 23.0 22.6 0.00 0.74 Taken from Run 93 179 30.4 29.3 o.00 0.97 315 44.2 53.3 0.01 1.44 Taken from Run 93 531 69.5 91.0 0.01 2.22 984 110 172 0.02 3.51 Taken from Run 93 1,450 148 257 0.03 4.71 1,570 157 279 0.03 5.01 1,700 168 303. 04 5.36 1,910 184 340 0.04 5.84 2,080 196 372 0.05 6.23 2,220 206 397 0.05 6.55 1 site 2,380 213 427 0.05 6.77 2 clusters of sites 2,530 215 456 o.o6 6.84 2,940 222 536 0.07 7.05 3,600 224 667 0.08 7.10 4,760 228 894 0.11 7.21 6,540 236 1,240 0.15 7.40 8,770 240 1,680 0.21 7.49 12,500 243 2,420 0.30 7.51 18,400 n52 3,580 0.44 7.65 24,700 263 4,820 0.59 7.85 32,200 278 6,290 0.77 8.13 42,300 298 8,280 1.o01 8.54 59,700 332 11,700 1.43 9.20 67,600 346 13,300 1.62 9.48 50,000 305 9,800 1.19 8.58 37,100 278 7,270 o.89 8.04 27,100 259 5,290 o.65 7.66 21,300 249 4,150 0.51 7.48 15,300 240 2,960 o.36 7.33 10,700 232 2,060 0.25 7.20 7,000 224 1,340 0.16 7.02 4,520 216 848 0.10 6.83 2,820 208 515 o.o6 6.60 2,220 202 398 0.05 6.44 2,120 199 380 0.05 6.33 3 active sites 2,010 193 358 o.o4 6.15 1 active site 1,900 184 338 o.o4 5.85 1,740 174 309 o.04 5.53 No sites 516 68.7 88.3 o.o01 2.19.

130 Run 108 Date 6/6/67 Surface CSF-1 Liquid H2 Pressurel7.1 psia Orientation Upwards Average Depth of Liquid 8 in. Convection Shield Yes qcnet x 103 qM x 103 qL x 103 A 1 Tcorr ATs (watts) (watts) (w(OK) (OK) Comments cm 22.9 3.93 4.3 0.00 0.15 54.1 7.86 9.1 0.00 0.30 85.3 11.8 14.5 0.01 0.45 135 17.0 23.2 0.01 o.65 212 24.6 37.0 0.01 0.94 1 site 238 27.2 41.5 0.01 1.03 Above site inactive —another site intermittently active 293 30.9 51.7 0.01 1.17 2 steady sites 348 34.8 61.7 0.01 1.32 426 39.8 76.3 0.01 1.51 Small patch 515 43.8 93.0 0.01 1.66 Additional sites and clusters 623 45.9 114 0.01 1.74 772 47.9 143 0.02 1.81 975 50.3 182 0.02 1.90 1,400 53.4 284 0.03 2.01 Sites cover ~40% of surface area 1,980 57.1 379.o05 2.13 2,480 60.8 476 o.o6 2.26 Sites cover -80% of surface area Surface appears to be covered 3,050 65.5 589 0.07 2.43 with sites 3,570 68.1 690 0.08 2.52 4,400 72.3 854 0.10 2.66 5,170 75.7 1,000 0.12 2.77 5,980 79.4 1,160 0.14 2.89 5,550 77.3 1,080 0.13 2.82 4,810 74.1 935 0.10 2.73 3,950 70.2 766 0.09 2.59 3,360 67.3 650.o08 2.49 2,800 63.7 538 0.07 2.36 2,170 60.3 417 0.05 2.25 1,620 56.3 309 0.04 2.11 i7200 52.9 226 0.03 1.99 823 48.7 153 0.02 1.84 540 42.2 98.2 0.01 1.60 ~20 active sites 397 36.2 71.1 0.01 1.37 296 29.9 52.5 0.01 1.13 221 24.6 38.7 0.00 0.94 8 active sites 195 22.5 34.1 0.00 o.86 173 20.7 30.0 0.00 0.79 136 17.0 23.4 0.00 o.65 2 active sites 122 16.5 20.8 0.00 o.63 110 15.2 18&7 0.00 0.58 2 active sites-very small bubbles 101 14. 117.1 0.00 0.55 1 active site 92.0 13.1 15.6 0.00 0.50 No active sites 64.o0 10.5 10.6 0.00 0.40 36.5 6.8 5.9 0.00 0.26

131 Run 111 Date 6/7/67 Surface CSF-1 Liquid H2 Pressurel7.3 psia Orientation Upwards Average Depth of Liquid 12 in. Convection Shield No qnet x 103 qM x 103 qL x 103 A x 3Tcorr ATs (watts) (watts) (watts) (OK) (OK) Comments cm 29.4 5.76 4.7 0.00 0.22 46.5 8.12 7.6 0.00 0.31 61.9 11.0 10.0 0.00 0.42 90.6 14.4 1500.00 0.55 134 19.9 22.5 0.00 0.76 198 26.7 33.8 0.00 1.02 248 32.0 42.5 0.01 1.21 319 38.8 55.3 0.01 1.47 3 well separated sites 286 35.6 49.4 0.01 1.35 2 sites 245 31.4 42.0 0.01 1.20 4 sites-all separate 229 29.3 39.3 0.00 1.12 4 sites 350 39.6 61.2 0.01 1.50 More individual sites 388 40.9 68.5 0.01 1.55 Clusters developing 425 41.7 75.6 0.01 1.58 463 43.2 82.7 0.01 1.64 515 44.0 92.8 0.01 1.67 Patches developing 584 45.3 106 0.01 1.72 649 46.1 119 0.01 1.75 718 48.2 132 0.02 1.82 793 49.0 147 0.02 1.85 875 50.8 163 0.02 1.92 970 52.4 181 0.02 1.98 1,090 54.2 204 0.02 2.05 1,190 55.5 224 0.03 2.09 Sites cover ~40% of surface area 1,310 57.6 247 0.03 2.17 1,490 58.7 282 0.03 2.21 1,620 6o.5 3o8 o.o4 2.27 1,810 62.9 345 0.04 2.36 1,980 64.5 379 0.05 2.41 2,240 65.8 429 0.05 2.46 Surface appears to be covered 2,780 69.7 535 0.07 2.59 with sites 3,560 74.9 687 o.08 2.78 4,390 78.9 851 0.10 2.91 5,030 82.3 975 0.12 3.02 5,970 86.2 1,160 0.14 3*15 5,530 84.6 1,070 0.13 3.10 4,690 81.2 908 0.11 2.99 3,860 77.3 745 0.09 2.86 3,050 72.6 587 0.07 2.70 2,350 67.6 450 0.05 2.53 2,070 65.2 396 0.05 2.44 1,780 62.9 338 0.04 2.36 1,450 60.0 274 0 05 2.26

132 Run 111 (Continued) q net x 103 qM x A Tcorr ATsomment (watts) (watts) (watts) (OK) (OK)mments cm2 1,080o 56.3 203 0.02 2.13 Site density decreasing uniformly over surface 857 52.7 159 0.02 1.99 689 49.0 126 0.02 1.85 548 45.3 99.1 0.01 1*72 443 41.4 79.2 0.01 1.57 ~30 active sites 345 37.7 60o.6 0.01 1.43 "20 active sites 268 33.0 46.3 0.01 1.26 12 active sites 230 29.6 39.4 0.00 1.13 8 active sites 192 26.2 32.6 0.00 1.00 4 active sites 173 23.8 29.4 0.00 0.91 3 active sites 152 21.5 25.7 0.00 0.82 No active sites 124 17.6 21.0 0.00 0.67 89.6 14.4 14.8 o.oo 0.55 66.7 11.0 10.9 o.oo 0.42 51.8 8.91 8.4 0.00 0.34 35.9 6.29 5.8 0.00 0.24

133 Run 112 Date 6/7/67 Surface CSF-1 Liquid H2 Pressurel7.1 lbf/in.2 Orientation Upward Average Depth of Liquid 7 in. Convection Shield No ne t x 103 qM x 103 qL x 103 A Tcorr ATComments (watts) (watts) (watts) (OK) (OK)mments cm2 23.2 5.24 3.5 0.00 0.20 Outer dewar rotated 90~ 36.4 7.07 5.7 o.oo 0.27 47.8 9.17 7.6 o0.00 O. 70.8 14.9 11.0 0.00 0.57 99.0 16.2 16.3 0.00 0.62 122 18.6 20.3 0.00 0.71 161 23.8 27.0 0.00 0.91 198 27.8 33.6 o.00 1.06 345 31.4 40.1 0.00 1.20 286 36.2 49.3 0.01 1.37 344 41.4 59.6 0.01 1.57 403 45.9 70.4 0.01 1.74 476 48.7 84.2 o.o1 1.85 549 48.2 98.7 0.01 1.83 672 49.5 123 0.01 1.88 764 50.8 141 0.02 1.92 893 51.9 166 0.02 1.96 1,060 54.0 198 0.02 2.04 1,250 54.8 236 0.03 2.06 1,420 56.9 268 0.03 2.14 1,620 59.2 308 o.o4 2.22 1,860 61.8 355 o.o4 2.32 2,160 65.o 414 0.05 2.43 2,880 70.2 555 0.07 2.61 3,770 75.7 729 0.09 2.81 4,790 81.2 929 0.11 2.99 5,960 86.7 1,160 0.14 3.17 5,260 83.6 1,020 0.12 3.07 4,190 78.6 811 0.10 2.90 3,280 7 3.4 633 o.o8 2.72 2,470 67.6 473 o.o6 2.53 2,070 64.2 396 0.05 2.40 1,840 62.6 351 o.o4 2.35 1,660 61.3 316 o.o4 2.30 1,490 59.7 282 0.03 2.25 1,340 57.6 254 0.03 2.17 1,140 55.5 213 0.03 2.09 971 53.2 181 0.02 2.01 824 51.1 152 0.02 1.93 685 48.7 126 0.02 1.84 540 45.6 97.4 o.01 1.73 418 42.7 74.1 o. l 1.62 343 38.5 60.o o.ol 1.46 314 36.9 54.5 o. ol 1 40 245 32.0 42.0 o 0. l01 1.21 193 26.7 32.8 0.o00 1.02

134' Run 112 (Continued) qnet x 103 qM x 103 qL x 10 A Tcorr ATs Comments (watts) (watts) (watts ) (OK) (OK)Comments cm 163 23.6 27.4 0.00 0.90 135 20.2 22.6 0.00 0.77 113 18.1 18.6 0.00 o. 69 89.8 14.7 14.8 0.00 0.-6 72.5 12.3 11.8 0.00 0.47 58.2 11.0 9.3 0.00 0.42 40.3 8.1 6.3 o.oo 0.31

135 Run 116 Date 6/29/67 Surface CTF-1 Liquid H2 Pressure 17.0 psia Orientation Upward Average Depth of Liquid 12 in. Convection Shield No net x 103 qMx 103 L x 103 A Tcorr AT Comments (watts) (watts) (watts) (OK) (OK)Comments cm 25.4 6.o4 5.82 0.01 0.22 39.1 8.39 6.07 0.02 0.31 52.1 10.5 8.21 0.02 0.38 1 site 56.9 11.4 8.98 0.02 0.42 64.3 12.6 10.2 0.03 o.46 69.6 13.4 11.1 0.03 o.49 76.2 14.4 12.2 0.03 0.52 83.4 15.4 13.4 o.o4 o.56 93.2 16.6 15.1 0.04 o.60 102 17.7 16.6 o.o4 o.64 114 19.1 18.6 o.o0 0.69 123 20.2 20.3 0.05 0.73 3 sites 135 21.6 22.3 o.o6 0.78 2 of the above sites inactive — 148 23.3 24.7 0.07 0.84 2 othe sites active 2 other sites active 179 26.7 30.1 0.08 0.95 216 30.4 6.5 0.10 1.0o8 6 sites 257 34.6 43.9 0.12 1.22 7 sites 308 38.7 53.0 0.14 1.35 12 sites 348 42.0 60.3 o.16 1.46 17 sites 423 46.9 74.2 0.20 1.61 489 50.5 86.4 0.23 1.72 613 54.6 110 0.29 1.81 Individual sites distributed 742 60.2 134 o.36 1.397 over entire surface 861 63.8 157 0.42 2.04 1,060 68.9 196 0.52 2.14 1,280 73.2 238 o.63 2.20 1,560 72.1 294 0.78 2.00 1,840 77.6 348 0.92 2.07 2,250 83.4 428 1.14 2.08 2,970 93.3 568 1.51 2.10 3,880 102 746 1.98 1.97 4,890 114 943 2.50 1.91 6,020 128 1,160 3.o8 1.86 5,270 117 1,020 2.70 1.80 4Y200 100 809 2.15 1.72 3,310 86.1 636 1.69 1.63 2,430 72.1 465 1.23 1.55

136 Run 116 (Continued) qInet x 103 qM x 103 IL x 103 A xTcorr ATs (watts) (watts) (watts ) (K) Comments cm 1,790 61.4 341 0.91 1.46 1,240 51.7 235 0.62 1.37 918 44.8 172 0.46 1.27 672 39.1 125 0.33 1.18 512 34.6 94.1 0.25 1.09 Small partical of something 383 29.6 0 0.1 o.96 0 29.6.9-7 0.19 0.96 in liquid moving around the 293 25.8 52.6 0.14 0.86 surface and acting as a site 244 23.3 43.5 0.12 79 Bubbles from sites are very 197 20.6 3407 0.09 0.70 small-almost appear to be a mist 161 18.4 28.1 0.07 o.64 Many sites 134 16.1 23.2 o.o6 o.56 111 14.3 19.1 0.05 0.50 89.9 12.2 15.3 0.04 o.43 76.2 10.7 12.9 0.03 0.38 -50 sites —small bubbles 65.3 9.67 11.0 0.03 0.34 54.1 8.43 9.01 0.02 0.30 44.2 7.45 7.25 0.02 0.27 33.9 6.13 5.48 0.01 0.22 ~40 sites 23.8 4.88 373. 0.01 0.18.30 sites

Run 120 Date 6/30/67 Surface CTF-1 Liquid H2 Pressure 17.1 psia Orientation Upward Average Depth of Liquid 9 in. Convection Shield No ne t x 103 M x 103 qL 103 A Tcorr ATx (watts) (watts) (watts) (OK) (OK)mments cm 0.0 0.0 0.0 0.00 0.00 Time - 0 2,022 104 378 1.00 2.97 0.5 min 2.94 1 2.79 2 2.73 3 2.61 5 2.49 7 2.42 11 2.35 15 2.27 20 2,035 82.0 385 1.02 2.11 35 2.05 66 2.00 81 1.94 130 2,036 78.1 386 1.03 1.92 144 1.90 166 1.90 177

138 Run 121 Date 6/30/67 Surface CTF-1 Liquid H2 Pressurel7.3 lbf/in.2 Orientation Upward Average Depth of Liquid 5 in. Convection Shield No qMx103 9x13 qnet x 103 X x i0 LxA Tcorr AT Comments (watts) (watts) (watts) (OK) (OK)Comments cm2 8,320 161 1,610 4.27 1.95 7,090 144 1,370 3.64 1.92 5,930 127 1,140 3.04 1.87 4,810 110 927 2.46 1.79 3,890 95.8 748 1.99 1.71 3,150 83.6 604 1.60 1.62 2,560 74.3 490 1.30 1.57 2,090 66.4 399 1.06 1.51 1,710 59.7 325 0.86 1.44 1,370 53.3 261 o.69 1.36 1,130 48.o 214 0.57 1.29 921 43.5 173 o.46 1.22 746 39.5 139 0.37 1.15 612 36.2 114 0.31 1.09 492 32.4 90.6 0.24 1.01 Many sites producing small bubbles 402 29.5 73.5 0.20 0.95 328 26.6 59.5 o.16 0.87 269 23.7 48.4 0.13 0.79 223 21.6 39.7 0.11 0.73 181 19.3 32.0 O. o8 o.66 151 17.3 26.3 0.07 o.60 123 15.4 21.3 o.o6 0.54 102 13 6 17.4 0.05 o.48 84.6 12.1 14.3 o.o04 o.43 68.4 10.4 11.4 0.03 0.37 55.8 9.15 9.21 0.02 0.33 ~50-60 sites 45.3 8.07 7.35 0.02 0.29 37.0 7.18 5.87 0.02 0.26 29.7 6.41 4.60 0.01 0.24 ~30-40 sites 24.1 5.69 3.63 0.01 0.21 ~30 sites 21.0 5.26 3.10 0.01 0.19 25 sites

139 Run 122 Date 7/5/67 Surface CSF-4 Liquid H2 Pressurel7.2 psia Orien-tationUpward Average Depth of Liquid 7 in. Convection Shield No Inet x 103 qM x 103 qL x 103 A 6Tcorr ATs (watts) (watts) (watts) (OK) (K)Comments cm2 20.9 4.63 3.20 0.00 0.18 26.4 5.67 4.09 0.00 0.22 31.9 6.54 5.00 0.00 0.25 38.9 7.31 6.23 0.00 0.28 47.4 8.44 7.68 0.00 0.32 58.2 10.0 9.49 0.00 0.38 70.5 12.1 11.5 0.00 0.46 85.1 13.7 14.1 0.00 0.52 101 15.8 16.8 0.00 o. 60o 122 18.3 20.5 0.00 0.70 148 21.4 25.0 0.00 0.82 178 24.3 30.4 0.00 0.93 214 28.2 36.7 0.00 1.07 259 32.8 44.5 0.00 1.25 313 37.8 54.3 0.00 1.44 376 43.6 65.6 0.00 1.66 454 50.6 79.5 0.01 1.92 394 3C6.5 107 0.01 1.38 Patch of 40-50 sites covering 594 56.5 107 0.01 1.38 663 39.0 123 0.01 1.48 794 37.4 149 0.01 1.42 Patch growing Second patch of 10-20 sites, 970 36.5 184 0.01 1.38 Most likely from runner of first patch. 1,190 37.4 226 0.01 1.41 1,440 38.6 276 0.02 1.45 1,750 39.4 338 0.02 1.48 2,180 40.7 423 0.03 1.52 Sites on.50% of surface 2,960 41.9 576 0.04 1.56 Surface almost coverered with 3,860 43.6 753 0.05 1.61 sites Surface appears to be covered 4,790 46.5 936 o.o6 1.72 with sites 5,920 50.2 1,160 0.07 1.86 7,110 53.5 1,390 0.09 1.95 8,320 56.7 1,630 0.10 2.06 7,720 53.9 1,510 0.10 1.96 6,400 48.5 1,250 o.o8 1.77 5,250 44.0 1,030 0.07 1.61 4,200 39.4 821 0.05 1.46

140 Run 122 (Continued) qnet x 103 qM x lO03 qL x 03 A Tcorr ATo s (watts) (watts) (watts) (OK) (OK)mments cm2 3,300 35.7 645 0.04 1.32 2,500 32.4 486 0.03 1.20 1 860 29.4 361 0.02 1.10 Site density decreasing uni1.370 -26. 265 0.02 1.00 formly formly 1,130 25.6 218 0.01 0.97 859 24.3 165 0.01 0.92 643 22.2 122 0.01 o.84 Many sites producing small 482 21.1 90.9 0.01 0.79 bubbles 371 20.0 69.2 0.00 0.76 286 18.9 52.7 0.00 0.72 -100 active sites 206 17.4 37.3 0.00 o.66 260-80 active sites distributed 165 16.3 29.4 0.00 0.62 over surface 130 15.1 22.6 0.00 o.58 99.8 13.8 17.0 0.00 0.53 ~40 active sites 76.1 12.0 12.6 0.00 0.46 ~25 active sites 56.o 10.5 8.98 0.00 0.40 16 active sites 45.9 9.30 7.23 0.00 0.36 10 active sites 37.2 8.26 5.71 0.00 0.32 4 active sites 31.3 7.14 4.77 0.00 0.27 3 active sites 26.5 6.58 3.93 0.00 0.25 1 active site 23.3 5.93 3.44 0.00 0.23 1 active site 20.7 5.54 3.00 0.00 0.21 1 active site 15.7 4.72 2.17 0.00 0.18 No sites

141 Run 125 Date7/11/67 Surface CSF-3 Liquid H2 Pressurel7.1 psia Orientation Upward Average Depth of Liquid 12 in. Convection Shield No -net x 103 qM X 103 qL x 103 A 5Tcorr ATs (watts) (watts) (watts (OK) (~K) Comments cm 20.4 5.63 2.91 0.00 0.21 26.1 6.71 3.82 0.00 0.26 31.7 7.75 4.73 0.00 0.30 38.7 9.13 5.83 0.00 0.35 47.6 10.6 7.30 o.oo0 o.40 57.8 11.8 9.o8 0.00 0.45 70.4 13.8 11.2 0.00 0.53 8.6 15.8 13.8 1.8 0 o.6o 105 18.1 17.1. oo0 o.69 127 20.4 21.0 0.00 0.78 Approximate initial vapor point — 154 23.8 25.6 0.00 0.91 exact point missed 187 26.7 31.7 0.00 1.02 228 30.5 38.9 0.00 1.16 274 33.0 47.5 0.00 1.26sites distributed over surface Individual sites and clusters f334 34.3 59.2 0.00 1.31 of 3 and 4 sites 407 35.1 73.3 0.00 1.34 494 35.1 90.5 0.01 1.33 607 35.6 113 0.01 1.35 737 35.1 138 0.01 1.33 892 35.6 169. o1 1.35 Individual sites and clusters 1, 160 37.2 221 0o.0o1 1.- 41 over the entire surfac 1,550 39.3 297 0.02 1.48 2,130 41.0 412 0.03 1.53 2,900 43.4 564. 04 1.6i 3,850 46.3 750 0.05 1.71 4,/80 49.2 933 o.o6 1.82.5,920 53.3 1,160 0.07 1.97 5,220 50.4 1, 020 0. 07 1.86 4,280 45.9 835 0.05 1.71 3,290 41.8 641 o.o4 1.55 2,530 38.5 491 0.03 1.44 1,780 34.7 343 0.02 1.30 1,570 32.2 263 0.02 1.21 1,050 30.5 202.o01 1.15

142 Run 125 (Continued) c net x 103 M x 103 qL 03 A 5Tcorr Tx Commenrts (watts) (watts) (watts) (OK) (OK) cm2 Site density and bubble size 780 28.0 148 0.01 1.o6 decreasing 608 26.3 115 0.01 0.99 470 24.2 88.o 0.01 0.91 356 22.1 65.9 0.00 o.84 268 19.5 49.0 0.00 0.74 194 17.4 34.9 0.00 o.66 153 15.5 27.2 0.00 0.59 122 13.9 21.2 0.00 0.53 94.3 12.2 16.2 0.00 0.47 ~100 active sites 70.6 10.3 11.9 0.00 0.39 53.8 8.66 8.89 0.00 0.33 ~40-50 active sites 41.0 7.57 6.59 0.00 0.29 -30 active sites 31.7 6.32 5.00 0.00 0.24 ~25 active sites 26.4 5.76 4.07 0.00 0.22 ~20 active sites 20.7 4.98 3.10 0.00 0.19 16 active sites 15.7 4.20 2.26 0.00 0.16 12 active sites

143 Run 134 Date 7/18/67 Surface CSF-4 Liquid N2 Pressure 17.2 lbf/in.2 Orientation Downward Average Depth of Liquid 8 in. Convection Shield Yes qnet x 103 qM x 103 qL x 103 A Tcorr ATsComments (watts) (watts) watts) (OK) (OK) cm2 20.6 4.87 3.10 0.00 0.56 48.3 10.2 7.52 0.00 1.17 89.4 17.8 14.1 0.00 2.04 19 36.2 30o.3 o.oo 4.16 Vapor formed sometime after 236 40.6 38.6 o.oo 4.67 4.67OK

144 Run 146 Date 7/20/67 Surface CSF-1 Liquid N2 Pressurel7.6 psia Orientation Downward Average Depth of Liquid 9 in. Convection Shield Yes ne t x 103 qM x 103 qL x 103 A Tcrr ATs Comments (watts) (watts) (watts) (OK) (OK) cm 21.9 4.40 3.45 0.00 0.51 54.8 10.9 8.67 o.o0 1.25 87.0 17.8 13.7 0.00 2.04 172 31.2 27.9 0.00 3.58 Formed vapor sometime after 3.58~K

145 Run 147 Date 7/21/67 Surface CSF-4 Liquid N2 Pressure 17.2 psia Orientation Upward Average Depth of Liquid 8 in. Convection Shield No net x 103 qM x 103 qL x 103 A XTcorr AT s (watts) (watts) (watts) (K) (K)Comments cm2 37.0 9.47 5.43 0.00 0.30 77.3 16.4 12.0 0.00 0.53 175 30.9 28.3 0.00 0.99 389 57.5 65.5 0.00 1.84 788 98.4 136 0.00 3.15 1,030 119 18o 0.00 3.81 1,280 141 224 0.00 4.53 1,560 137 281 0.00 4.40 Patch covering about 5% of 2,030 138 372 0.00 4.42 surface. 2,830 140 530 0.00 4.48 Patch spreading. Patch covers about 65% of 4,520 148 861 0.0o 4.74 surface. Surface almost covered with 7,540 160 1,460 0.00 5.12 active sites 12,600 176 2,460 0.01 5.62 18,500 192 3,610o 0.01 6.14 27,700 211 5,410 0.02 6.74 42,900 242 8,420 0.015 7.72 Temperatures might not have reached steady state values 62,600 278 12,300 0.04 8.88 as readings were rushed 50,500 259 9,910 0.03 8.26 39,400 240 7,730 0.02 7.67 30,900 225 6,050 0.02 7.19 23,700 212 - 4,630 0.01 6.76 17,700 199 3,450 0.01 6.38 13,000 187 2,530 0.01 6.oo 8,300 173 1,600 0.01 5.55 5,270 160 1,010 0.00 5.12 3,850 152 730 0.00 4.88 2,650 145 494 0.00 4.66 1,840 140 336 0.00 4.48 -50 active sites 1,140 121 201 0.00 3.88 ~25 active sites 909 105 158 0.00 3.38 ~15 active sites 698 90.1 120 0.00 2.89 1 steady, 5-6 intermittent sites 567 74.4 97.2 0.00 2.38 No sites 1253 23.8 19.6 0.00 0.76

146 Run 151 Date 7/31/67 Surface CF-2 Liquid H2 Pressurel7.2 psia Orientation Upward Average Depth of Liquid 11 in. Convection Shield No x qL x A XTcorr ATs (watts) (watts) (watts (OK) (OK)ments cm2 90.5 15.1 14.9 0.00 0.57 109 17.4 18.0 0.00 0.66 1 intermittent site 129 13.7 22.8 0.00 0.52 Patch of -10 sites, above site inactive. 166 12.8 30.2 0.00 0o49 Patch growing, site density in patch not great. 225 13.3 41.8 0.00 0.51 554 15.7 106 0.00 0.60 Patch covers -25% of surface. 989 17.5 192 0.00 0.67 Patch covers -60% of surface. 2,010 21.3 393 0.00 0.81 Patch cover -85% of surface. Surface appears to be covered 3,310 25.9 647 0.00 o.98 with sites. 4,050 27.8 794 0.00.o06 4,780 28.6 938 0.01 1.09 5,580 29.0 1,100 0.01 1.10 4,870 26.7 955 0.01 1.01 3,800 24.6 745 0.00 0.94 3,o60 22.9 599 0.00 0.87 2,330 21.2 455 0.00 0.81 Site density decreasing. 1,930 20.1 377 0.00 0.77 1,580 19.2 308 0.00 0.73 1,270 18.2 247 0.00 o.69 1,020 17.4 198 0.00 o.66 Site density uniform over surface 798 16.5 154 0.00 o.63 648 15.8 125 0.00 o.60 527 15.1 101 0.00 0.58 429 14.5 81.8 0.00 0.55 341 14.0 64.5 0.00 0.53 271 13.6 50.7 0.00 0.52 215 12.7 39.8 0.00 0.48 177 12.2 32.4 0.00 o.46 1100 sites, large bubbles, 145 11.5 26~3 0.00 0.44 low frequency. 117 10.9 21.0 0.00 0.42 95.9 10.1 16.9 0.00 0.39 79.7 9.81 13.8 0.00 0.37 64.2 9.34 10.8 0.00 0.36 ~40-50 intermittent sites 52.7 8.74 8.68 0.00 0.33 ~30 intermittent sites 43.5 8.18 6.97 0.00 0.31 2 steady sites, ~25 intermit355. 7.44 5.54 0.00 0.28 tent sites 28.5 6.67 4.31 0.00 0.25 22.2 5.93 3.20 0.00 0.23 1 steady site

Run 154 Date 8/1/78 Surface CSF-1 Liquid H2 Pressure17.7 psia Orientation Vertical Average Depth of Liquid 11 in. Convection Shield No qnet x 103 qM x 103 qL x 103 A 1 Tcorr ATs Comments (watts) (watts) watts) (OK) (OK)mments cm2 20.3 3.64 3.28 0.00 0.11 29.5 5.34 4.77 0.00 0.17 59.9 10.4 9.77 0.00 0.37 139 21.4 23.2 0.00 0.78 162 23.4 27.4 0.00 0.86 194 27.0 32.9 0.00 1.00 220 28.7 37.7 0.00 1.o06 246 32.4 42.2 0.01 1.20 297 38.9 50.9 0.01 1.45 375 46.5 64.9 0.01 1.74 1 site near top of surface. 495 52.9 87.2 0.01 1.98 5-6 sites on top 1/3 of surface. 661 54.4 120 0.01 2.03 Sites on top 1/2 of surface. 824 47.7 153 0.02 180 1 site at very bottom of surface, sites in triangle above it. i,o60 53- 7 198 0.02 2.02 1,330 58.0 250 0.03 2.18 1,730 64.6 329 0.04 2.43 Sites cover ~60% of surface. 2,280 72.0 435 0.05 2.69 2,950 76.9 567 0.07 2.87 Sites cover -80% of surface. 3,770 86.3 726 0.09 3.21 4,840 94.9 937 0.11 3.51 Sites appear to cover entire 6,230 109 1,200.15 4.03 surface 8,290 117 1,610 0.20 4.26 6,530 103 1,270 0.15 3.76 5,360 91.6 1,040 0.13 3.37 4,280 81.1 828 0.10 2.99 About 2 mm along bottom. 3,~60 70.4 649' o8'6 Almost void of sites. 2,570 61.5 495 0.0o6 2.29 1,900 53.7 363 o.o4 2.00 1,350 45.3 258 0.03 1.70 969 39.3 183 0.02 1.48 675 32.8 127 0.02 1.23 Sites in triangle with 600 angle, 503 27.8 93.7 0.01 1.05 located at bottom of surface. 347 24.5 63.5 0.01 0.93 3 areas with sites 241 20.8 43.5 0.01 0.79 110 steady sites on top 196 18.6 35.0 o. O 0.71 1/2 of surface

148 Run 154 (Continued) qnet x 103 qM X 103 qL X 103 A Tcorr ATs Comments (watts) (watts) (watts (OK) (OK) cm 162 17.0 28.6 0.00 o.64 132 14.0 23.4 0.00 0.53 7 steady sites 108 13.1 18.8 0.00 0.50 6 steady sites 89.8 12.1 15.3 0.00.46 4 steady sites, all on top 1/2 of surface. 72.8 10.5 12.3 0.00 0.40 3 steady sites 62.4 9.28 10.5 0.00 0.35 3 steady sites 49.8 7.93 3.25 0.00 0.30 2 steady sites 43.3 7.30 7.11 0.00 0.28 2 steady sites 35.0 6.11 5.69 0.00 0.23 No sites

149 Run 158 Date 8/3/67 Surface CF-4 Liquid H2 Pressure 17.7 psia Orientation Upward Average Depth of Liquid 10 in. Convection Shield No 103 qL x 103'net x 103 qM x 0 q x 0 A 1 Tcorr ATs (watts) (watts) (watts) (OK) (OK) cm 22.1 0.974 4.16 0.00 0.11 36.1 2.90 6.56 0.00 0.18 60.7 4.72 11.0 0.00 0.30 91.4 7.89 16.5 0.00 0.40 Patch of 7-8 sites. 106 10.4 18.8 0.00 0.40 146 10.4 26.8 0.00 0.39 ~20 sites in patch. 241 10.3 45.5 0.00 0.40 O10% of surface covered. 471 10.4 90.9 0.00 0.40 700 10.6 136 0.00 0.40 ~35% of surface covered. 1,290 10.6 253 0.00 0.49 ~60% of surface covered. 2: 3500 12.9 450 ~0~~ o6o0 Patch appears to cover the 2,500 12.9 4'o 0.00 o.6o entire surface. 3,120 15.9 613 0.00 0.69 4,470 18.1 878 0.01 0.84 6,000 22.2 1,180 0.01 1.02 7,650 27.0 1,500 0.01 1.21 6,810 31.9 1,340 0.01 1.13 5,590 29.7 1,100 0.01 o.98 4,590 25.9 899 0.01 o.83 3,760 22.0 767 0. 00 0.74 3,o8o 19.4 603 0.00 o.65 Site density very large. 2,490 17.1 487 0.00 o.56 2,o040o 14.8 400 0.00 0.50 1 660 13.1 324 0.00 o.45 1,360 11.9 265 0.00 0.42 1,110 11.0 217 0.00 0.39 905 10.1 177 0.00 0.36 726 9.52 141 0.00 0.34 Many sites, bubble size 587 8.83 114 0.00 0.32 getting smaller. 484 8.35 93.8 0.00 0.30 397 7.88 76.7 0.00 0.29 317 7.54 61.1 0.00 0.27 253 7.19 48.5 0.00 0.27 207 6.97 39.4 0.00 0.25 170 6.67 32.3 0.00 0.24 140 6.32 26.4 0.22~0. ~~1500sites distributed over surface. 113 5.72 21.1 0.00 0.21 90.3 5.59 16.7 0 0.20 100 active sites. 74.2 5.28 13.6 0.00 0.19 50.8 5.11 9.02 0.00 0.18 50-60 active sites.

150 Run 165 Date 8/7/67 Surface CSF-3 Liquid H2 Pressurel7.1 psia Orientation Vertical Average Depth of Liquid 9 in. Convection Shield No -qnet x 103 XM x 103 qL x 13 A 3Tcorr ATs (watts) (watts) (watts) (OK) (OK)mments cm 21.8 7.10 2.90 0.00 0.27 45.1 11.7 6.58 0.00 0.45 94.4 19.1 14.9 0.00 0.73 104 20.5 16.5 0.00 0.78 113 21.3 18.1 0.00 o.81 124 22.6 20.0 0.00 o.86 137 24.3 22.2 0.00 0.93 152 25.5 25.0 0.00 0.97 1 site ~1/2 way from top 167 26.8 27.6 0.00 1.02 of surface. 187 28.4 31.2 0.00 1.08 197 28.8 33-1 0.00 1.10 4 sites above previous site. 243 31.3 41.8 0.00 1.19 2 additional sites 1/4 way from top, sites near top. 1 site near bottom center, 4_8 29.7 80.6 O.O1 1.12 * sites in triangle above- it. 833 35.4 157 0.01 1.34 Surface -30% covered with sites. 1.390 40.7 267 0.02 1.53 Surface -50%0 covered with sites. 2,130 45.1 411 0.03 1.69 Surface ~65% covered with sites. 2,910 48.8 565 0.04 1.82 Surface -80% covered with sites. 3,980 52.0 774 o.05 1.93 Surface appears to be covered 575 1,102 0.07 2.15 with sites. 7,700 61.8 1,507 0.10 2.26 6,160 48.0 1,205 0.08 1.75 5,070 43.9 992 o.o6 1.61 3,960 39.1 774 0.05 1.44 3,280 35.4 641 0.04 1.31 Surface still covered with sties. 2,680 31.7 522 0.03 1.18 Site density decreasing, 2,220 29.2 433 O.053.08 especially near bottom. 1,860 26.3 362 0.02 0.98 1,490 23.9 290 0.02 0.89 1,210 21.8 234 0.01 0.82 Few sites on bottom 2mm of 976 19.3 189 0.01 0.72 surfa ce.

Run 165 (Continued) qne t x 103 qM x 103 qL x 103 A 5Tcorr ATs Coents (watts) (watts) (watts) (OK) (OK)Comments cm 791 17.7 152 0.01 o.66 Few sites on bottom 5mm of 635 15.8 122 0.01 0.59 surface. 526 14.6 101 0.01 0.55 428 13.4 81.7 0.01 0.50 333 12.2 63,2 0.00 0.47 Few sites on bottom 6 mm 267 11.0 50.5 0.00 0.42 of surface. 220 10.2 41.5 0.00 0.39 179 9.48 33.4 0.00 0.36 143 8.62 26.5 0.00 0.33 114 8.07 21.0 0.00 0.31 76.3 7.26 13.6 0.00 0.28 -15 sites, mostly 1/2 way 57.,0 6.84 9.89 0.00 0.26 from top, 57.0 from top. 40.4 6.50 6.68 0.00 0.25 ~10 steady sites. 32.4 5.90 5.22 0.00 0.23 5 steady sites. 22.0 5.34 3.29 0.00 0.20 3 steady sites.

152 Run 169 Date 8/10/67 Surface CSF-4a Liquid H2 Pressure 17.2 psia Orientation Vertical Average Depth of Liquid 13 in. Convection Shield No x 103 qL x 03 q-net x 103 x 103 L x 103 A Tcorr AT Comments (watts) (watts) tts (OK) (OK)Comments cm 22.2 4.72 3.44 0.00 0.18 34.9 6.73 5.55 0.00 0.26 59.4 1o6 9.62 0.00 0.41 93.3 15.2 15.4 0.00 0.58 125 18.0 21.0 0.00 0.69 139 127 248 0 0 8-10 sites on same side of sur15~9 12.7 24.8 0.00~ 0. 49 facet bottom ones 1/3 way up. 181 14.0 32.9 0.00 0.53 298 15.6 Sites in same area cover 2985155.7 0.00 20% of surface. 522 17.9 99.4 0.01 0.67 912 19.9 176 0.01 0.75 Sites cover 3o50% of surface. 1,450 22.1 282 0.02 0.82 Sites cover ~60% of surface. 2,260 25.2 442 0.03 0.93 Sites cover.95% of surface. 35320 35.0 648 0.04 1.29 Sites appear to cover surface. 3,920 37.8 765 0.05 1.39 5,270 30.9 1,030 0.07 1.11 7,240 34.6 1,420 0.09 1.24 6,420 33.3 1,260 0.08 1.19 5,290 30.9 1,040 0.07 1.11 4,470 28.8 876 o.o6 1.04 3,490 26.3 684 o.o4.96 2,910 24.7 570 0.04 0.90 2,210 22.1 431 0.03 0o.81 1,700 20.2 332 0.02 0.75 1,180 17.9 230 0.01 0.67 838 16.3 162 0.01 0.61 588 14.8 113 0.01 0.55 Few sites on bottom 2mm 396 13.6 75.4 0.00 0.52 260 12.4 48.8 0.00 0.47 184 11.4 34.0 0.00 0.43 Most sites on top 1/3 of surface. 139 10.4 25.4 0.00 o.40 ~75 sites 102 9.96 18.2 0.00 0.38 -50 sites 68.9 9.07 11.8 0.00 0.35 ~25 sites 55.3 8.51 9.23 0.00 0.32 ~ 15 sites 45.4 8.04 8.79 0.00 0.31 10 sites 37.4 7.52 7.23 0.00 0.29 5-7 sites 22.4 5.66 4.31 0.00 0.22 3 steady sites.

REFERENCES 1, Westwater, J, W. "Boiling of Liquids,11 T. B. Drew and J. W. Hoopes, Jr. Ed., Advances in Chemical Engineering, 1, Academic Press Inc., 1956. 2. Rohsenow, W. M., "Heat Transfer with Boiling," W. M. Rohsenow Ed., Developments in Heat Transfer, Press, 1964. 3. Leppert, C. and Pitts, C. C., "Boiling," T. F. Irvine, Jr. and J. P. Hartnett Ed., Advances in Heat Transfer, 1, Academic Press Inc., 1964. 4. Zuber, N., "Recent Trends in Boiling Heat Transfer Research, Part I: Nucleate Pool Boiling," Applied Mechanics Review, i No. 9, p. 663, 1964. 5. Bankoff, S. G., "Ebullition from Solid Surfaces in the Absence of a Pre-existing Gaseous Phase," Trans,. ASM, I79 p. 7359 May, 1957. 6. Clark, H. B.,, Strenge, P. S., and Westwater, J. W., "Active Sites for Nucleate Boiling," Chemical Engineering Progress Symposium Series, 55 No, 29, p. 103Y 1959. 70 Corty, C. and Foust, A. S., "Surface Variables in Nucleate Boiling," Chemical Engineering Progress Symposium Series, 59, No. 17, p. 1, 1955. 8. Griffith, P. and Wallis, J. D,, "The Role of Surface Condition in Nucleate Bo:iling," Chemical Engineering Progress Symposium Series, 56, No. 30, p. 499 1960. 9. Hsu, Y. Y. and Graham, R. W., "An Analytical and Experimental Study of the Thermal Boundary Layer and Ebullition Cycle in Nucleat Boiling," NASA TN D-594 (:1961). 10. Hsu, Y. Y., "On the Size Range of Active Nucleation Cavities on a Heating Surface," Trans. ASME, Series C, J. of Heat Transfer, 85, p. 207, 1963. 11. Bergles, A. E. and Rohsenow, W. M., "The Determination of Forced-Convection Surface-Boiling Heat Transfer," Trans. ASME, Series C, J. of Heat Transfer. 86, No. 3, p. 3659 Aug. 1964. 12. Han, C, Y. and Griffith, Po,, "The Mechanism of Heat Transfer in Nucleate Pool Boiling, Part I: Bubble Initiation, Growth and Departure," Int. J. Heat Mass Transfer, 8, No. 6, p. 887, June 1965. 153

154 REFERENCES (Continued) 13. Howell, J. R. and Siegel, R., "Incipience, Growth and Detachment of Boiling Bubbles in Saturated Water from Artificial Nucleation Sites of Known Geometry and Size,," Presented at the Third International Heat Transfer Conference, Chicago, Ill., Aug. 7-12, 1966. 14. Bankoff, S. G., "Entrapment of Gas in the Spreading of a Liquid Over a Rough Surface," AIChE Jour. 4, No. 1, p. 24, March 1958. 15. Marto, P. J. and Rohsenow, W. M., "Effects of Surface Conditions on Nucleate Pool Boiling of Sodium," Trans. ASME, Series C, J. of Heat Transfer, 88, No. 2, p. 196, May 1966. 16. Hatton, A. P. and Hall, I. S., "Photographic Study of Boiling on Prepared Surfaces," Presented at the Third International Heat Transfer Conference, Chicago, I.., Aug. 7-12, 1966. 17. Merte, H. Jr. and Clark, J. A., 1"Pool Boiling in an Accelerating System," Trans. ASME, Series C, J. of Heat Transfer, 8, p. 233, Aug. 1961. 18. Raben, I. A., Beaubouef, R. T., and Commerford, G., "A Study of Nucleate Pool Boiling of Water at Low Pressure," Chemical Engineering Progress Symposium Series, 61, No. 57, p. 249, 1965. 19. Gaertner, R. F., "Distribution of Active Sites in the Nucleate Boiling of Liquids,," Chemical Engineering Progress Symposium Series, ~5_ No. 41, p. 52, 1963. 20. Berenson, P. J., "Experiments on Pool-Boiling Heat Transfer," Int. J. Heat Mass Transfer, 5, p. 985, 1962. 21. Kurihara, H. M. and Myers, J. E. "The Effect of Superheat and Surface Roughness on Boiling Coefficients," AIChE Jour., 6, No. 1, p. 83, March 1960. 22. Lyon, D. N., "Boiling Heat Transfer and Peak Nucleation Boiling Fluxes in Saturated Liquid Helium Between the \ and Critical Temperatures," K. D. Timmerhaus, Ed., International Advances in Cryogenic Engineering) 10, Plenum Press, p. 371, 1965. 23. Githinji, P. M, and Sabersky, R. H., "Some Effects of Orientation of the Heating Surface in Nucleate Boiling," Trans. ASME, Series C, J. of Heat Transfer, 85, No. 4, p.o 379, Nov. 1963.

155 REFERENCES (Continued) 24. Marcus, B. D. and Dropkin, "The Effect of Surface Configuration on Nucleate Boiling Heat Transfer," Int. J. Heat Mass Transfer, 6, p. 6839 1963. 25. Richards, R. J,, Steward, W. G., and Jacobs, R. B., "A Survey of the Literature on Heat Transfer from Solid Surfaces to Cyrogenic Fluids," NBS TN-122 1961. 26. Mulford, Ro N, and Nigon, J. P., "Heat Exchange Between a Copper Surface and Liquid Hydrogen and Nitrogen,?" Los Alamos Scientific Laboratory, LA-1416, 1952. 27. Class, C, R,, DeHann, J, R,, Piccone, M,, and Cost, R, B., "Boiling Heat Transfer to Liquid Hydrogen from Flat Surfaces," K, D. Timmerhaus, Advances in Cryogenic Liquids, 5, Plenum Press, p. 254, 1960. 28. Class, C. R., DeHann, J. R., Piccone, M,, and Cost, R. B., "Pool Boiling Heat Transfer to a Cryogenic Liquid," WADC Technical Report 58-528, 1958. 29. Drayer, D, E. and Timmerhaus, K. D., "An Experimental Investigation of the Individual Boiling and Condensing Heat Transfer Coefficients for Hydrogen," K, D. Timmerhaus, Ed,, Advances in Cryogenic Engineering, 7, Plenum Press, p. 401, 1962, 30. Graham, R, W., Hendricks, R. C., and Ehlers, R. C., "An Experimental Study of the Pool Heating of Liquid Hydrogen in the Subcritical and Supercritica"l Pressure Regimes over a Range of Accelerations," K. D. Timmerhaus, Ed., International Advances in Cryogenic Engineering, 10, Plenum Press, p. 342, 1965. 31. Fredrickson, G, O. and Schweikle, J. D., "Thermo and Hydrodynamic Experiment Research Module in OrbitsFinal Report," Douglas Missile and Space Systems Division, DAC-60594, March 1967. 32, Sherley, J, E., "Nucleate Boiling Heat Transfer Data for Liquid Hydro~ gen at Standard and Zero Gravity," K. Do Timmerhaus, Ed., Advances in Cryogenic Engineering, 8 Plenum Press, p. 495, 1963. 335 Steinle, H. F, "Revliew of Zero-G Studies Performed at General Dynamics/ Astronautics "Advances in the Astronautilcal Sci.ences 14 p. 95. 34. Tusk, G., "Zero-G Report —LFH2 Boiling Threshold," GD/A Report 55D 85935 May 1962.

156 REFERENCES (Continued) 35. Drayer, D. E., "Nucleate Boiling of Hydrogen, " I and EC Fundamentals, 4, No. 2, p. 167, May 1965. 36. Zuber, N. and Fried, E., "Two-Phase Flow and Boiling Heat Transfer to Cryogenic Liquids," Americal Rocket Society Jour., 32, No. 9, P. 1332, Sept. 1962. 37. Forster, H. K, and Zuber, N,, "Dynamics of Vapor Bubbles and Boiling Heat Transfer " AIChE Jour. 1, No. 4, p. 531, Dec. 1955. 38. Forster, H. K. and Greif, R., "Heat Transfer to a Boiling LiquidMechanism and Correlations," Trans. ASME, Series C, J. of Heat Transfer, 81, No. 1, p. 42, Feb. 1959. 39. Cryder, D. S. and Gilliland, E. R,, Ind. Eng. Chem., 24, p. 1382, 1932. 40. Hord, J., Jacobs, R. B., Robinson, C. C., and Sparks, L. L., "Nucleation Characteristics of Static Liquid Nitrogen and Liquid Hydrogen," Trans ASME, Series A, J. of Engineering for Power 86, No. 4, p. 485, 1964. 41. Good, R. J. and Ferry, G. Vo, "The Wetting of Solids by Liquid Hydrogen," K. D. Timmerhaus, Ed., Advances in Cryogenic Engineering, 8, Plenum Press, p. 306, 1963. 42. Brentari, E. Go and Smith, R. V., "Nucleate and Film Pool Foiling Design Correlations for 02, H2, N29 and He," K. D. Timmerhausg Ed., International Advances in Cryogenic Engineering, 10 Plenum Press, p. 325, 1965. 43. McAdams, W. H., Heat Transmission, McGraw-Hill Book Co., 1954. 44. Scott, R. B,, Cryogenic Engineering, D. Van Nostrand Co., Inc., 1959. 45. Powell, R, L. and Bunch, Nat. Bur. of Standards Publ., R-188. 46. Feller, W., Probability Theory and Its Applications, 1, John Wiley and Sons, Inc., New York, 1950. 47. Corruccini, R. J., "Properties of Liquid Hydrogen," Nat. Bur. of Standards Publication, R412, 19650 48. Scott, R. B,, Denton, W. H., and Nicholls, C. M., Technology and Uses of Liquid Hydrogen, Pergamon Press, 1964,

157 REFERENCES (Concluded) 49. Strobridge, T. R., "The Thermodynamic Properties of Nitrogen from 64 to 300~K Between 0.1 and 200 Atmospheres," Nat. Bur of Standards, TN129, 1962. 50. Handbook of Chemistry and Physics, 40th ed., Chemical Rub'ber Publishing Co,, Cleveland, Ohio, 1959. 51. Kreith, F.,, Principles of Heat Transfer, 2nd ed., International Textbook Co., Scranton, Pa., 1966. 52. Goodwin, R. D,, Diller, D. E., Roder, H. M,, and Weber, L. A., "The Densities of Saturated Liquid Hydrogen," Nat. Bur, of Standards Publication, R226, 1961L

9ZZt LZ8Z0 9L06 0 11111111111fll I IIIIIIIIIIIII1111l111F NV91HOII. 10AIIS:3AINl