YT Froon ’E* § 1 she Ly |3 ;’nr\o'fl i . ; Sty E\ s Js.c\. k{ Adfiizfflzf‘l}cY PUCUMENY bOLLEZQiAfi% ARIETTA ENERGY SYSTEMS LIBRARIES ' ’: I". s,‘\l"'( :f ’[ ‘% (TRIRAE LA 3 4456 OD349BA7? 5 - ORNL 1777 Recctors-Research and Power =2 o L FUSED SALT HEAT TRANSFER 'PART Il; FORCED CONVECTION HEAT TRANSFER IN CIRCULAR TUBES - CONTAINING NaF ~KF-LiF EUTECTIC H W. Hoffmcn J. Lones - CENTRAL RESEARCH LIBRARY DOCUMENT COLLECTION LIBRARY LOAN COPY DO NOT TRANSFER TO ANOTHER PERSON If you wish someone else to see this document, send in name with document and the library will arrange a ioan. CARBIDE A . A DIVISIO nc;m anmcz upu | .!;, Tznuzaszfi Beagfiorflmeauh ‘aad Power This document consists of pages. Copy _g2 of 5_____15 copies. Series A. Contract No. W-7L405, eng. 26 Reactor Experimental Engineering Division FUSED SALT HEAT TRANSFER PART II: FORCED CONVECTION HEAT TRANSFER IN CIRCULAR TUBES CONTAINING NeF-KF-LiF EUTECTIC H. W. Hoffman J. Lones Date Issued: FEB 1 1955 OAK RIDGE NATIONAL IABORATORY operated by CARBIDE AND CARBON CHEMICALS COMPARY Union Carbide and Carbon Corporation Post Office Box P Oak Ridge, Tennessee MARTIN MARIETTA ENERGY SYSTEMS LIBRARIES . HEARDRIRAET 3 445L D349L87 & ol 1. Biology Library ho, 2-3. Central Research Library 50. i, Health Physics Library 51. 5. Reactor Experimental 52. - Engineering Library 53. 6-13. Laboratory Records Department 54, 1k, Laboratory Records, ORNL R.C. 554 15. G. M. Adamson 56. 16. R. G. Affel 57 . 17. L. G. Alexander 58. 18. C. J. Barton 59. 19. S. E. Beall 60. 20. E. S. Bettis 61. 21. D. S. Billington 62. 22. F. Blankenship 63. 23. E. P. Blizard 64, 24, E. G. Bohlmann 65. 25. J. 0. Bradfute 66. 26. R. W, Bussard 67. 27. A. D. Callihan 68. 28. D. W, Cardwell 69, 29. C. E., Center T0. 30. R. A. Charpie T1l. 31. G. A. Christy 12. 32. H. C. Clalborne T3. 33. G. H. Clewett Th, 34, W. G. Cobb T5. 35. S. I. Cohen ‘ T76. 36. C. D. Coughlen TTo 37 S. J. Cromer 78. 38. F. L. Culler T9. 39, M. C. Edlund 80. 40. J. L. English 81. 41. J. Y. Estabrook 82. k2, A. P. Fraas 83. 43. W. J. Fretague 8h. Lh, J, H. Frye, Jr. 85. k5. €. B. Graham 86. 46, N, D. Greene 87, k7. J. C. Greiss 88. 8. W. R. Grimes -1i- INTERNAL DISTRIBUTION EXTERNAL DISTRIRUTTON - 89. AF ifiggggpresentative, Burbank 0. AF P t Representative, Seattle 91l. AF Plant Representative, Wood-Ridge 92. American Machine and Foundry Company 93. ANP Project Office, Fort Worth Reactors~Research P. ORNL 17T&R » d- Péwer Ce. Hamilton - 0. Harms W. Hoffman R. Johnson H. Jordan R. Kasten W. Keilholtz Kertesz 5. Kitzes F. Lansing E. Lynch N. Ly'en D, Manly B. McDonald R. Mann A. Mann L. Meem - C. Miller M. Nessle D. Palmer W. Parkinson ¥. Poppendiek D. Powers T. Robinson W. Rosenthal F. Salmon W. Savage D. Shipley Sisman P. Smith A. Swartout H. Taylor G. Thomas Tobias M. Warde M. Welnberg A. Lan&i. o E. Winters C. Zmola ORNL Document Reference Library}>Yei2 Branch - #:- a ooy ok-10k. 105. 106. 107-111. 112. 113 . 11k, 115-120. 121-123, 124, 125, 126. 127. 128. 129-130. 131. 132. 133-136. 137. 138. 139, 140, 14134k, 1ks, 146-153. 154, 155, - 156. 157 "’16{’- 161-162. 163. 16k, 165, 166. 167. 168. 169-170. 171. 172, 173-17h. 175. 176. 177. 178. ~179. 180"186 - 187. 188. 189. ~1ii- Argonne Netlonal Laboratory Armed Forces Special Weapons Project (Sandis) Armed Forces Special Weapons Project, Washington Atomic Energy Commission, Washington Babcock and Wilcox Company Battelle Memorial Institute Bendix Aviation Corporation Bettis Plant (WAPD) Brookhaven National Iaborsatory Buresu of Ships Bureau of Yards and Docks Chicago Operations Office Chicego Patent Group Chief of Naval Research Commonwealth Edison Company Department of the Navy - Op-362 Detroit Edison Company duPont Company, Augusta duPont Company, Wilmington Duquesne Light Company Fngineer Research and Development Laboratories Foster Wheeler Corporation General Electric Company (ANPD) General Electric Company (APS) General Electric Company, Richland Henford Operastions Office Towa State College Kaiser Engineers Knolls Atomlc Power Laboratory Los Alamos Scientific Iaboratory Massachusetts Institute of Technology (Benedict) Materials Iaboratory (WADC) Monsanto Chemical Company Mound Laboratory National Advisory Committee for Aeronautics, Cleveland National Advisory Committee for Aeronautics, Washington Naval Research lLaboratory Newport News Shipbuilding and Dry Dock Company New York Operations Office North American Aviation, Inc. Nuclear Development Associates, Inc. Nuclear Metals, Inc. Pacific Northwest Power Group Patent Branch, Washington Pennsylvania Power and Light Company Phillips Petroleum Company (NRTS) Powerplant Laboratory (WADC) Pratt & Whitney Aircraft Division San Franciso Operations Office (Fox Project) “o . . 190. Sylvania Electric Products, Inc. 191. Tennessee Valley Authority (Dean) 192. USAF Headquarters 193. USAF Project Rand 194, U. S. Naval Postgraduate School 195. U. 5. Naval Radiological Defense Laboratory 196-197. University of California Radiation laboratory, Berkeley 198-199. University of California Radiation Laboratory, Livermore 200. Vitro Englineering Division 201. Vitro Leboratories 202, Walter Kidde Nuclear Laboratories, Inc. 203. Westinghouse Electric Corporation (IAPG) 204-218. Technical Information Service, Oak Ridge 219. Division of Research and Medicine, AEC, ORO -V- TABLE OF CONTENTS INTRODUCTION=-=--mem e mmemmme e e m e e DESCRIPTION OF APPARATUS-=-=s--voremmmcmn— SYSTEM CALIBRATION=--=--mmmvecmccmmnc e ———— ANAIYSIS OF DATAccc-vmmmm e memcecc e e e e e DISCUSSION-mmm~mmmmmmmmmmmmmmmmm e -- CONCLUSIONS=~werrommmmrcc e e e - APPENDIX A - Nomenclabureee---=-scecmmm—wa-= APPENDIX B - Bibliography-=------=--- ———— APPENDIX C - Physical Properties of Flinak P e e L R T I R e D WS Wm e m Sue WS Y M A G e S i A Sk SEy M A W S W S W S MR SR TR e e W SR me S we - e e e ey e Rek AT U R A TR R W S R RO M T g SR e e S e e B el et AU ME ML W S e W e W R e P W N KR N R R - e e ow W S M W o S S R G W T W G e e i e S N M YE P M M R W NS g M TR Ep e e e e v = ey e ik G I Ve SR D AR S N T S s e W G R S S e BN WD NS G ST WS W S W M e W e S £ i2 1k 22 30 L1 Ll -1- SUMMARY An experimental determination has been made of the heat transfer coef- ficients for the eutectic mixture of sodium, potassium and lithium fluoride (Flinak) flowing in forced convection through circular tubes. Heat, electrically generated in the tube wall, was transferred uniformly to the fluid during passage through a small diameter tube. Iong tubes (x/d >100) of nickel, Inconel and 316 stainless steel were used. The varisbles involved covered the following ranges: Reynolds modulus : 2300 - 9500 Prandtl modulus : 1.6 - 4.0 Average fluid temperatures: 980 - 1370°F 2 Heat flux : 9,000 - 192,000 Btu/hr-ft It was found that forced convection heat transfer with Flinak can be represented by the general correlation for heat transfer with ordinary fluids (0.5& Npr < 100). The existence of an interfacial resistance in the Flinak-Inconel system has been established and its compogition determined. Preliminary measurements of the thermal conductivity and thickness of the film have been made. The results verify the effect of this film on Flinak heat transfer in small di- ameter Inconel tubes. Thermal entry lengths, determined from the variation of the local heat transfer coefficients in the entrance of the heated section, have been corre- lated with the Peclet modulus. Introduction Fused salts have potential application as heat transfer media wherever heat must be transferred to, or removed from, a system at temperatures above 600°F. Because the vapor pressures of fused salts are generally low at these elevated temperatures, operation at atmospheric or near atmospheric pressures is possible. In a reactor, fused salte possess a further advantage. By com- bining a number of salts, a fluid can be designed such that it can be used as a moderator asg well as a cqolant. If uranium is present this fluid can also be the fuel. This report presents the second part of a continuing investigation into the heat transfer characteristics of fused salts of interest to the ANP Project. First reported (5)* were the results of an experimental determination of heat transfer coefficients for molten sodium hydroxide. It was concluded that, with respect to heat transfer, sodium hydroxide may be classified as an ordi- nary fluid (0.5< Npy <100). This report describes heat transfer with the eutectic mixture of sodium, potassium and lithium fluoride (11.5-42.0-46.5 mole %) known as Flinak. Some results on Flinak heat transfer were reported in an earlier memorandum (6). Flinak melts at approximately 850°F and is temperature stable to above lBOOoF. At 1500°F the viscosity of Flinak is about three times the viscosity of water at 60°F. In a reactor Flinak can be used as a coolant or in combination with *¥ Reference 5 is to be considered as Part I of the series on fused salt heat transfer. o | 3 UF3 or UFy as a fuel-coolant. The efficiency of a fluid as & coolant can be defined as the work required per Btu of heat removal. Poppendiek, Rosenthal and Burnett, in a report now in preparation, have evaluated this efficiency for a number of coolants (sodium, lithium, bismuth, sodium hydroxide and Flinak) using four temperature difference parameters. In general, they conclude that Flinak is a good coolant, the best being of course lithium. Description of Apparatus The experimental system designed to measure the heat transfer coefficient of molten Flinsk flowing in tubes was similar in all major respects to that previously described (5) for molten sodium hydroxide heat transfer studies. It consisted of two tanks; one of which rested on a scale to enable fluid flow rate measurements, and a test section located between the two tanks. Flow through the test section was effected by means of gas pressure applied above the fluid in the tanks. Several modifications were made to provide an experi- mental system possessing increased flexibility and incorporating metals compatible with molten Flinak at high temperatures. TFigures 1 and 2 present several views of the experimental system. Nickel will contain Flinak but is subject to fatigue failure at tempera- tures above 1000°F. Therefore, Inconel, which exhibits good corrésion re- sistance to Flinak and retains its structural strength at elevated temperatures, was adopted as the material of construction. One of the major inconveniences encountered with the sodium hydroxide experimental system was the inability to easily replace system components. To g SUM i P TANK Fig M. G eneral View of Experiment al S ystem VOLTAGE TAP POWER LEAD SUPPORT BEARING I EAR L ] POWER LEAD THERMOCOUPLES TEST SECTION —/ MIXING, PO MIXING POT ARD HEALTER i = ’_'..'-' . =i . — e e T Tt Fig. 2. Test Unit—Shown Partially Assembled. eliminate this difficulty a number of changes were made for the Flinak system: (1) The tanks were redesigned to incorporate flanges with metal O-ring seals for introducing thermowells, dip lines, etc., into the top of the tanks. (2) Each tank was contained in an open-topped stainless steel Jacket. The heaters were attached to this jacket so that tank replacement could be accomplished without complete removal of heaters and thermal insulstion. (3) The test section was connected to the system by Swage-lck unions. Use of these connectors, in which a positive seal is effected partly by compression of the ferrule and partly by forcing the nose of the ferrule into the tube wall, facilitated the interchange of test units. At high temperatures, thermal expansion gives rise to stresses which are sufficient to” cause buckling of the test section if the tube ends are held rigidly{ To alleviate this condition, the scale used in the weight-rate determination was oriented such that the wheels were "in-line" with the direction of thermal expansion of the system. A turn-buckle arrangement allowed the weigh tank to be moved so as to compensate for the expansion of the test section and the connecting lines. To further eliminate strain in the test section, the power terminals were supported on roller bearings. This vielded & test section which was undistorted. The test section consisted of a 24 inch length of small diameter tubing which was heated by the passage of an electric current through the tube wall. Power was supplied by a transformer rated at U Kva and was introduced to the test section through copper terminals silver-soldered to the tube at each end. In the course of the experiment tubes which were made of nickel, Inconel, and 316 stainless steel were used as the test section. The dimensions of these tubes are given in the following table: TABLE 1 Qutside Wall Length Diameter Thickness Dlameter (d) Material (inches) (inches) Nickel 0.1875 0.035 204 Inconel 0.225 0.025 137 316 s.s. 0.250 0.035 133 With the Inconel and 316 s.s. test sections, flow at a Reynolds modulus of 10,000 could be attained. As a result of'the temperature limitation on the nickel test section, a maximum Reynolds modulus of only 6000 was possible. In each case over 99% of the heat generation occurred in the tube wall. The temperature of the outer surface of the test section was measured with’thermocouples which were resistance welded to the tube surface. The fluid mixed-mean temperatures were measured in mixing pots at eachhend of the test section. Details of the mixing pot and‘test section end are shown in Figure 3. The voltage impressed on the test section was obtained at 8 locations along the tube. At each position one of the wires of the tube surface thermo- couple at that point was used as the voltage probe. The system contained 350 pounds of Flinak, charged to the system as a well-mixed powder and melted while alternately evacuating the region above the melt and then purging and mixing the fluid with helium. The helium was dried and 6xygen removed by bubbling it through a tank filled with sodium- potassium alloy. UNCLASSIFIED ORNL-LR-DWG 3448 THERMOCOUPLE THERMOCOUPLE THERMOCOUPLE POWER TERMINAL FLUID MIXING POT THERMOCOUPLE WELL MIXING POT POWER TERMINAL THERMOCOUPLE Fig. 3. Test Section and Mixing Pot Detail . -8- i . -9- System Calibration To insure the reliability of the data obtained it is necessary to cali~ brate the various measuring devices associated with the system. It had been determined in the sodium hydroxide experiment that the readings of the scale used in the weight-rate measurements were consistent and accurate and that the current flowing in the tube wall had no effect on the readings of the tube surface thermocouples. Since the factors affecting these were unchanged by apparatus modifications, no additional check was made for the Flinak experiment. Similarly, no preliminary tests were made to determine the approach to equili- brium of the test unit during the time available for an experimental run. However, it was found during system operation that equilibrium conditions were attained well before the end of each run. The thermocouples used in measuring the fluid mixed-mean temperatures were calibrated at the freezing points of lead, zinc and aluminum. Table II shows the results obtained from the calibration of the fluid mixed-mean thermo- couples used in a typical test series. TABLE II Calibration of Fluid Mixed-Mean Thermocouples Temperature Thermocouple ~ OF 21 - 23 ok 621.1 -0.06% -0.08 -0.08 -0.06 787-1 -0.06 «0.07 -0.07 -0.06 1219.5 ~0.055 -0.06 -0.07 -0.055 ¥Body of table gives the correction (in millivolts) to be added to the observed reading to obtain the true reading. . ‘” " ~10- Since the tube surface thermocouples were formed on the tube after the test unit had been assembled, calibration of these couples was difficult. Therefore, the couples were checked in two ways. Samples of the wires used were taken, formed into thermocouples and calibrated in a tube furnace with reference to a standard couple. The results showed a discrepancy of approxi- mately 1°F between the observed and true readingé at 1200°F. On the completion of a test series, the tube was removed and sectioned. Each small section, with thermocouple attached, was then calibrated‘at the temperatures corre- sponding to the freezing points of lead, zinc and aluminum. For purposes of comparison, the tube surface thermocouples on a dummy test section were checked similarly. Table III gives the result of this calibration check for 4 couples (different in each case) at the aluminum point. TABLE III Calibration of Tube Surface Thermocouples (Melting Point of Aluminum = 27.44% millivolts) I-1% I-2 Thermocouple (millivolts) (millivolts) 1 27.43 27. bh 2 27.46 27.k2 3 27.45 27.43 L 27. 4k 27 .42 * I-1 TInconel dummy test section. I~-2 Inconel test section used in run series G. A It is seen that the output of the tube surface thermocouples remained essentially unchanged even after operation at temperatures up to lh5OQF and despite a great deal of temperature cycling. Part of the total energy put into the test section is lost to the system environment. Since the magnitude of this loss must be known to obtain a heat balance for the system, a heat loss calibration was made. The technique employed has been described in Part I (5). As the system heat loss is a function of the density and distribution of the insulation around the test section and of the end conditions, it was necessary to obtain a heat calibration for each test unit. Figure 4 shows the outside tube surface temperature for various power levels without fluid flow. The dip in the curves at the position corresponding to the tube center probably results from the tempera- ture profile in the thermal insulation caused by the mixing pot and power terminal guard heaters at the test section ends. The system heat loss curves for the several test units are given in Figure 5, where At 1s the average outside tube surface temperature, tw,ave- minus the temperature of the system environment, te. Analysis of Data The local coefficient of heat transfer (or local conductance) is defined by the equation A)x (Qf/ (l) W R pe—— (tg - ty)x _Zy_ OUTSIDE TUBE SURFACE TEMPERATURE (°F) UNCLASSIFIED ORNL-LR-DWG 3449 1800 1600 O T \D\ 1400 | | N\ { —0 ~ 0 e e - {I-——-"-D\ 1200 1000 \ S A A 2 N 800 ELECTRICAL HEAT INPUT (Btu/hr) O 444 600 A 337 0O 300 ® 250 A 204 400 | 200 0 0 2 4 6 8 10 12 14 S 18 20 22 24 DISTANCE FROM ENTRANCE OF TEST SECTION (in.) Fig. 4. Axial Profiles of Outside Tube Surface Temperatures for Varying Heat Input with no Fluid Flow. UNCLASSIFIED ORNL—LR—DWG 3450 450 400 A NICKEL NO.4 O INCONEL NO.1 / @® INCONEL NO.2 350 A NICKEL NO.2 300 250 200 q, (BTu/hr) 150 100 50 / 7 A ‘/ 200 400 600 800 1000 1200 1400 Uy ave = %) (°F) Fig. 5. System Heat Loss. 1600 ) -1k- where qf/A is the heat flux through the wall-fluid interface, tg, the temperature of the surface at this interface, tp, the fluid mixed-mean tempera- ture and the subscript, x, indicates the geometrical position on the surface at which the heat flux and temperature difference are evaluated. For large values of x (beyond the thermal entrance region) hy reaches a limiting value, h, which is the heat transfer coefficient for the region of both thermally and hydrodynamically established turbulent flow. In the correlations which follow the coefficient h is used unless otherwise noted. The inside surface temperature of thevtuhe is calculated from the measured outside tube surface temperature by the equation¥ 2 2 2 T r,~ -r by ~tg = M (r" In ¥ _ WS where r is the tube radius, the subscripts w and s refer to the outside and inside tube surfaces, respectively, and k; is the trermal conductivity of the tube metal. The source term, W, is given by w=3_-”_l_‘3r_E£ (3) where E is the voltage impressed on the test section, I, the current passing through the tube wall, and V, the volume of metal in the tube wall of the test ¥ The derivation of this equation is given in Reference (5). £ -15- section. Equation 2 is a modification of the simple conduction equation to allow for the generation of heat in the tube wall. The derivation assumes uniform heat generation in the tube wall, no longitudinal heat flow, and no heat losses through the surface, w. Figure 6 shows, for a typical test condition, that the voltage impressed on the test section is a linear function of the distance along the tube. Thus, the assumption of uniform heat generation in the derivation of eqpafiion 2 is seen to be reasonable. With uniform heat generation and the small axial temperature rise of the fluid, the fluid mixed-mean temperature at any point, X, along the test section can be obtained from the straight line drawn between tm,i and tm,o, the fluid mixed-mean inlet and outlet temperatures, respectively. The total heat generation in the tube wall is determined from the measured values of the voltage drop across the test section and the current passing through the tube wall. Part of the total heat goes to exfiernal loss and the rest into heating the fluid. The external loss is obtained from the curves of Figfire 5 for the given set of experimental data. The heat transferred to the fluid in passing through the test section is given by the equation 4 = W Cp (tm,o - ty, i) (&) where w is the fluid flow rate in lbs/hr. The heat transfer area is taken as the inside surface area of the test section between the two power terminals. The fieat flux is based on the heat gained by the fluid as given by equation L. The heat balance for each run is shown in Teble IV. Figure 7 shows a typical temperature profile for the outside tube wall. VOLTS UNCLASSIFIED ORNL—LR—DWG 34514 . 8 10 12 14 16 DISTANCE FROM ENTRANCE OF TEST SECTION (in.) Fig.6. Voltage Impressed On Test Section—RunI-1 20 22 24 TEMPERATURE (°F) 1440 1420 1400 1380 1360 1340 1320 1300 UNCLASSIFIED ORNL-LR-DWG 3452 RUN J-1 —--.\ /—" _——-O_——_ —9 FLUID MIXED-MEAN __-———"——- 4 6 8 10 12 14 16 18 20 22 24 DISTANCE FROM ENTRANCE OF TEST SECTION (in) Fig. 7. Typical Temperature Profiles Along Axis of Electrically Heated Test Section. ' -18- The physical properties of Flinak necessary to these calculations are presented graphically in Appendix C. Since both the axial and radial témperature differences were small, the physical properties were evaluated at the average fluid mixed-mean temperature. By applying similarity considerations to the differential equations for heat transfer in a moving fluid, it can be shown that the parameters déscribing forced convection heat transfer are related by the equation My = B (Nges Npy) (5) or Ngy = Do (Me, Npy) (6) where Ny, the Nusselt modulus = hd/k, Ngt, the Stanton modulus = h/cp G NRe, the Reynolds modulus =z d4G/u , and Npy, the Prandtl modulus = cp M/k. This functional relationship has also been predicted by the analogies com=- - paring heat and momentum transfer within a fluid flowing turbulently in a long tube (1, 9, 12, 13). Analysis of experimental turbulent forced convection heat transfer data shows that, in equation 5, the Reynolds modulus generally appears to the 0.8 power and the Prandtl modulus to the 0.4 power. The simplified empirical equation proposed by McAdams (1), | Nyy = 0.023 Nge0'8 mp 0% (7) has been found to correlate, to within + 20%, all data for the fluids termed olin -19- "ordinary", (0.5lOO) where fully developed turbulent conditions existed. The results covered the transition flow region and checked generally the correlation for ordinary fluids. An attempt to obtain data in the region of fully developed turbulent flow by raising the temperature above 1000°F was unsuccessful due to failure in fatigue of the nickel tube. The data demonstrated thét variations in the entrance conditions, arising from slight differences in geometry between the two ends of the test section, strongly affect transition region heat transfer. Thus, the results for flow through ORNL—LR—!WG 3453 REYNOLDS MODULUS Fig. 8. FLINAK Heat Transfer. 0.010 A NICKEL TUBE NO.1 v NICKEL TUBE NO.4 O INCONEL TUBE NO.4 @® INCONEL TUBE NO.2 ) 0O TYPE 316 STAINLESS STEEL TUBE & < Il > Ll Z'- \ o o %: &) 3 —-0.2 % /VST- NPr = 0.023 NRe ) L = sl @ 0.002 a ® 3 ® 0.001 1000 10,000 20,000 50,000 TABLE IV 4 EXPERIMENTAL DATA AND CAILCUIATED RESULTS - FLINAK HEAT TRANSFER Run Qr/A W tm, o~ ,if twsave tym.avel Ts-tm h Nre | Npyr Ng J Heat Balance Btu%/;r-f# 1bs/hr OFfjm OF o oF | Btu/hr-r£2-°F 5t (ar+ a5 /2e) F-1 23,215 L06.8 7.8 1005 1001 6.9 3364 2459 | 3.7} .00138 | .00331 0.95 F-2 36,920 431.h 11.7 996 981 15.5 2381 2428 | 4.0 | .000923| .00233 0.90 F-3 L2,284 431. 4 13.h4 1010 990 19.8 2136 2516| 3.9 | .000828] .0020k 1.02 F-l 96,241 365.k4 36.0 1116 109k 20.2 W76k 2968 | 2.8 | .00218 | .00430 0.91 F-5 95,591 372.6 35.8 1133 1108 21.9 4456 31471 2.7 | .00200 | .00384 0.93 F-6 129,893 543.0 32.7 1144 1123 16.4 7920 Y7691 2.6 | .0024k4 § .00L57 0.97 G-1 9,138 372.0 5.0 1238 | 1228 9.9 923 2779 1 2.0 | .00092 | .001khT 1.02 G-2 60,317 660.0 18.6 1230 1195 28.5 2116 46261 2.2 | .00119 | .00198 1.02 G=-3 119,170 780.0 31.1 1334 1279 L6.5 2563 64111 1.8 | .00122 | .00183 0.92 G4 121,397 86L.0 28.6 1291 1235 45.2 2686 6566 | 2.0 { .00115 | .00181 0.9k G~5 12k, 421 732.0 34h.6 1309 1245 5k, 3 2291 5661 | 2.0 | .00116 | .00181 0.98 G-6 125,993 306.0 28.73 1350 1297 h3.5 2896 7762{ 1.8 | .00119 [.00179}- 0©0.98 G=T 126,233 798.0 32.2 1340 1280 50.6 2495 63871 1.8 | .00116 | .0017TH 0©.99 G-8 133,606 1050.0 £25.9 1311 1258 1.7 3204 83371 1.9 | .00113 {00174 1.00 G-9 134,661 972.0 28.2 1299 1247 45.0 2992 75761 1.9 | .0011k4 | .00177 1.01 G-10{ 171,201 720.0 L8. 4 1331 1241 76.1 2250 5611 | 1.9 | .00116 {.00180 0.98 . H-1 121,747 796.8 31.1 1273 121k 50.0 2435 57961 2.1 ] .00113 | .00184 0.98 H-2 121,867 768.0 32.3 1269 1210 50.3 2L23 55521 2.1 | .00117 |.00191 1.01 I-1 52,957 338. 4 21.3 1329 1315 12.0 4413 44081 1.7 1 .00218 | .00314 0.96 I-2 66,5420 331.8 27.3 1341 1321 17.6 I77h L3s7 1 1.7 1 .00190 |.00272 0.90 I-3 133,849 Loo.6 43.% 1367 133h 28.1 4754 56381 1.7 | .00189 |.00267 0.96 I-4 191,559 369.0 70.8 1364 1353 34.5 5552 5076 | 1.6 | .00252 | .00350 0.96 J-1 80,170 733.2 22.9 1373 1350 16.1 4980 65861 1.6 | .00267 {.00371 1.02 J-2 98,864 870.0 23.8 1374 [ 1348 16.6 5956 78141 1.6 | .00269 |.00373] 0.9k T3 102,562 1027.8 20.9 1397 1371 15.7 6532 oh8L | 1.6 | .00250 |.00341 0.96 T -4 103,997 1027.8 21.6- | ~1399 | 1373 15.8 6709 95361 1.6 | .00256 |.00348 0.99 T -5 112,435 T7h4.6 30.L4 1384 1353 20.3 5525 69961 1.6 | .00280 |.00387 0.96 -6 112,541 810.0 29.1 1352 1332 20.7 5437 69811 1.7 | .0026L ].00377 1.05 =7 121,040 759.0 33.4 1391 1356 22.3 5428 68541 1.6 | .00281 |.00389 0.89 —'[a- ’ -22- the test section in one direction are indicated by the normal open triangles of Figure 8, while the inverted open triangles are for flow in the opposite direction. To sustain higher operating temperatures so as to obtain turbulent flow data, the nickel tube was replaced by an Inconel section and a second series of runs performed. The results of this test series (G) are shown in Figure 8 by the open circles. The data extended through the transition region into the regime of fully developed turbulent flow. It was apparent that the Flinak heat transfer was considerably lower (by a factor of two) than the heat transfer for sodium hydroxide or other ordinary fluids. Since the Prandtl modulus for Flinak lies in the same range as that for sodium hydroxide, o¢ne would expect the Flinak-Inconel data to fall on the curve correlating heat transfer for the ordinary fluids. In addition, these data do not agree with those of Test F obtaiped for Flinak flowing in a nickel tube. There appeared to be three possible explanations for the observed differences: (1) The experimental measurements were in error (2) The physical property data used in the analysis were incorrect (3) "Non-wetting" occurs or an interfacial resistance (film) existed at the metal surface From equation 1, which defines the heat transfer coefficient, it was seen that to account for the discrepancy on the basis of experimental error required that either the thermal flux or the temperature difference be in error by a factor of two or that errors in the proper direction exist in both . -23- quantities. Excellent heat balances eliminated the possibility of error in the thermal flux. While it was unlikely that a consistent error could occur in all nineteen of the tube surface thermocouples, the calibration of these couples was checked. No significant deviation from the expected calibration was observed. Similarly, the calibration of the thermocouples used in measuring the mixed-mean temperatures was verified. An analysis of the mixing- pots indicated that the mixed-mean fluid temperatures could be in error due to heat conduction losses from the thermocouple junctions. Although the probability was low that these losses were large enough to account for the variation in the data (an error of 10 to 20°F being required), the mixing- pots were redesigned to insure that the thermocouple lead wires remeined in a longer constant temperature region. A change in the thermal conductivity of Flinak from 2.6 to 0.9 Btu/hr-ft-°F or a change in viscosity from 11 to 0.02 1bs /hr-ft or a combi- nation of these two would be necessary if the data are to fit the accepted correlation. Recent physical property data for Flinak (thermal conductivity and viscosity) indicate the values used to be substantially correct. The existence of a "non-wetting' condition was eliminated by evidence showing that Flinak readily wets Inconel surfaces. However, the possibility of a high-melting film, caused by a reaction between Flinak and Inconel, forming on the inside tube surface presented itself. A section of the tube used in Test G was cut longitudinally, and a deposit was observed on the ingide tube surface. However, some question as to the origin of this film arose, since the tube had been exposed to air subsequent to its exposure to Flinak. - -2h- Therefore, a third series of runs (Test H) were made using a new Inconel test section and the modified mixing-pots. Only two points were obtained before shut-down resulting from failure of a system component. These are shown in Figure 8 by the closed circles and are sufficient to check the data of Test G. It was concluded that the difference between the cobserved and expected heat transfer was definitely not due to errors in the measurement of the mixed-mean temperatures. The test section from this series was cut longitudinally; and a green deposit, extending over the entire inside surface of the tube, was observed. Petrographic and electron diffraction examination of the film showed the major constituent to be K3CrFg. Some LigCrFg and small amounts of the oxides of chromium and nickel were also present. The latter were probably present at the stert of the runs since no pretreatment of the tube surfaces to remove metal oxides was attempted. K3CrF6 has a cubie structure, is green in color, and melts at a temperature of lOSSOC. With the film present, an apparent coefficient of heat transfer is given by the equation, h' = L (9) (te-tw) _ y (ag/A7) kg where A' is the inside surface area of the film, y, the film thickness and ki, the thermal conductivity of the film. Since the films are thin, the ratio of the heat transfer ares of the film to that of the clean tube, A'/A, is approximately one. Hence, (qe/A') in equation 5 may be replaced by (aqp/A). Thus, equation 5 can be written g h' o1 (10) From the Flinak-Inconel data, a value of 0.0002 hr-ft2-°F/Btu was ob- tained for the thermal resistance (y/ki) in equation 10. To check this, measurements were made of the thickness of the film on the Inconel surface and of the thermal conductivity of K3CrF6, The film thickness, determined from a photomicrograph (Figure 9) of one tube specimen, was found to be approximately 0.4 mils. A preliminary value of 0.133 Btu/hr-rt-°F was ob- tained (8) for the thermal conductivity over the temperature range 100-200°F. Using these values, a thermal resistance of 0.00025 hr-f+<-°F/Btu was ob- tained. This compared closely with the value (0.0002) deduced from the heat transfer data. At the same time an experiment was performed to determine if the current flowing through the tube wall of the test section influenced the film formation. Pieces of Inconel and nickel tubifig were allowed to remain in a still pot of molten Flinak for 24 hours. On removal a green film was observed on the Inconel tube. It was noted that the nickel tube showed no film under similar conditions. An additional check on the validity of the initial Flinak-nickel dafia wasg made. The results (Test I) are shown by the closed triangles of Figure 8. It was seen that the data were in general agreement with those of Test F. Again the influence of small geometrical variations at the test section entrance has been accentuated in the transition flow region. T FILM—y=0.4 mil INCONEL TUBE WALL Fig.9. Photomicrograph of Incone!l Tube Exposed to FLINAK. Magnification =250 X. L -21- Preliminary tests had indicated that Flinak could be contained for short times in 316 stainless steel without serious corrosion‘or the formation of insoluble fluoride deposits. Therefore, a 316 stainless steel test section was installed and a series of runs made (Test J). The results are shown by the open squares in Figure 8. As can be seen, Flinak in the absence of the interfacial film behaves, with respect to heat transfer, as an ordinary fluid; and conversely, the reduction in heat transfer in small diameter Inconel tubes can be attributed directly to deposits of the type, K3CrF6. Some informetion on the thermal entrance length was obtained from the experimental data. The entrance system consisted of a thermal entrance region preceded by a hydrodynamic entrance region of 8_to 13 tube diameters. In this investigation, (x/d)e was taken as the position at which the local heat transfer coefficient had decreased to within 10% of its fully established value. Figure 10 shows the local heat transfer coefficient as a function of x/d for a typical experimental run. Figure 11 presents, for both Flinak and sodium hydroxide, the observed veriation of (x/d)e with the Peclet modulus, (Npe = Npy * Npe)- As in the sodium hydroxide experiment, it is estimated that the maximum error in the heat transfer coefficients obtained for Flinak is T.S%. Discussion The first reported measurements for fused salt heat transfer were those of Kirst, Nagle and Castner (10) for the mixture, NaNO2-NaNO3-KNOg (40-7-53 weight percent), known as "HIS." They employed s 6 foot length of .--&.i LOCAL COEFFICIENT (Btu/hr-ft2-°F) Ny, 7000 6000 5000 4000 3000 2000 1000 ORNL- !!-!!E 3454 RUN F-2 Npe= 2428 \ Np, = 4.0 \ 20 40 60 80 100 120 140 160 180 200 x/d, DISTANCE FROM ENTRANCE OF TEST SECTION Fig. 10. Local Heat Transfer Coefficients for FLINAK Flowing in a Pipe under Uniform Flux Conditions. UNCLASSIFIED ORNL—LR—DWG 3455 50 A FLINAK ® SODIUM HYDROXIDE / / ® ® / @ o ,® y 20 / ® ® {x d)e +. 10 // // A\/ 5 / A/ A/ 3 6,000 10,000 50,000 N,., PECLET MODULUS Pe " Fig. 11. Thermal Entrance Length. 100,000 - | -30- 3/8 inch iron pipe which was heated by passage of an alternating electric current throfigh the pipe wall. The fluid was circulated through the pipe, and the inlet and outlet temperatures of the molten salt were measured with chromel-alumel thermocouples in wells at the pipe ehds. The outside surface temperatures of the pipe were measured at four positions along the pipe. Iacking data on the thermal conductivity of "HTS," they correlated their .k experimental results in terms of the function, hdébL . The average line through their experimental values was given by the equation 1.14 hd . aG ——-“O. = 0.0004k42 (-FZ.) (11) ovef thé Reynolds mbdulqs range 2000 to 30,000. Figure 12 shows the data of Kirst, Naglé and Castner in terms of the Colburn j-function using the preliminary values of Hoffman and Claiborne (7) for the thermal conductivity of "HTS." It is seen that considerable scatter exists in the data. Part I (5) of the current invéstigation reported measurements of the heat transfer coefficients for molten sodium hydroxide flowing turbulently in a nickel tube, 3/16" 0.D. x 0.035" wall thickness. The results are shown in Figure 13 and can be correlated by the equation =0. 4 NN‘u ° NPI‘ - 0.021 NReOQS (12) over the Reynolds modulus range 6000 to 12,000. Grele and Gideon (3) also obtained heat transfer data for sodium hydroxide flowing in an electricaliy heated tube. They used an Inconel tube having an outside diameter of 3/8" and a wall thickness of 1/16". A oy UNCLASSIFIED CRNL—LR-DWG 3456 Fig.12."HTS" Heat Transfer. 0.010 > & O ¥ 0.005 O— = O EEB O — T .” ‘// "--...,____‘\( O O j -~ O O 5 4 o O O — = 0. Re = /' C) C) C) z Q/ O = : / 0.002 o / Q DATA OF KIRST, NAGLE AND KASTNER (10) 0.00¢ 1000 2000 5000 10,000 20,000 50,000 Npge» REYNOLDS MODULUS 100,000 -3¢ - 0.020 0.040 2/ /VS‘r NPr 3 0.005 COLBURN FUNGTION, / 0.002 0.001 UNCLASSIFIED ORNL-LR-DWG 3457 ® HOFFMAN (5) { A GRELE AND GIDEON (3)—-HEATING A GRELE AND GIDEON(3)— COOLING A N pAlD A Aph A L, A Al A o A A A AL % ] . v - ° ‘K\ =0.023n._ O° / o Re 1000 2000 5000 10,000 20,000 50,000 Neo» REYNOLDS MODULUS Fig. 13. Sodium Hydroxide Heat Transfer. 100,000 L 33 centrifugal sump pump was used to circulate the molten sodium hydroxide through the system. The inlet and outlet fluid fiemperatures were measured in baffled mixing tanks at each end of the test section. The outside tube surface temperatures were measured with 1% chromel-alumel thermocouples located along the test section. In addition to heating data, Grele and Gideon made measurements on the cooling of sodium hydroxide in an air-cooled double-tube heat exchanger. Their results, for heating and cocling, are shown in Figure 13. Using the same system, Grele and Gideon (4) reported measurements on Flinak heat transfér (2000 € NRe «.20,000) in an Inconel tube. Their results corroborated the preliminary measurements of Hoffman (6) on Flinak, tests G and H, and are shown in Figure 1h. Recently, Salmon (14) using a double-tube heat exchanger, with sodium- potassium alloy (NaK) as coolant, measured heat transfer coefficients for the fluoride salt mixtuie NaF-ZrF),-UF), (50-46-4 mole percent) in a 0.269" I.D. nickel tube with a length to diameter ratio of L40. Center-line tempera- tures were measured at both the inlet and ocutlet of the fluid streams. An adjustable probe was used to meagure the surface temperatures on the annulus side of the center tube. The heat transfer coefficient was obtained by analysis of the data using the measured surfaée tefiperature and also by the graphical analysis techniéue suggested by Wilson (15). The results, recalculated from Salmon's data as the Colburn j-function, are shown in Figure 15. _.bg_ 2/ Ngi* Moy COLBURN FUNCTION, J ORNL-L!—! 3458 0.020 0.040 O GRELE AND GIDEON (4) ® HOFFMAN (6) 0.005 /-j=0.023 Nge 2 /./ -.______< / \ /7 o /4 O 0.002 O— 0O 00 1% b%e O ® O O OC O O O Qoo 00O o O 0.004 1000 2000 5000 10,000 20,000 50,000 Nge, REYNOLDS MQODULUS Fig. 14. FLINAK Heat Transfer. 100,000 2 Nsy* Vo 73 COLBURN FUNCTION, / omu& 3459 0.020 ’ 1 DATA OF SALMON (14) O CALCULATED FROM MEASURED SURFACE TEMPERATURES A CALCULATED BY WILSON LINE ANALYSIS 0.010 0.005 —"10 O 4 . — Q M\ - /=10.023 NReO.Z i 2 ‘ O / 0.002 £ 0.001 1000 2000 5000 10,000 20,000 50,000 Ngpe, REYNOLDS MODULUS Fig.15. NaF~ZrF,~UF, {50—46—4 mole %) Heat Transfer. 100,000 e 36- The results of this report show that Flinak, when suitably contained, behaves ag an ordinary fluid as far as heat transfer is concerned. However, in Inconel a reaction at the fluid-metal interface results in an insoluble, high-melting film which considerably reduces Flinak heat transfer. The effect of this on a reactor, where a fixed amount of heat must be removed in the heat exchanger, is readily apparent. Consider, the fuel to NaK heat exchanger degigned for a 50 megawatt eirculating fuel reactor. In this heat exchanger the fuel is circulated on the outside of Inconel tubes (3/16" 0.D. x 0.017" wall thickness) through which NaK is flowing. The Reynolds modulus of the fuel is 5000 and it enters the heat exchanger at 1500°F and exits at 1100°F. If the fuel is the mixture NaF-LiF-KF-UF) (11-45-L41-3 mole %), then the fuel- side heat transfer coefficient is approximately 5000 Btu/hr-ft2-°F. For the design conditions, the NaK (coolant-side) coefficient is 18,500 Btu/hr-fte-oF. The total thermal resistance is given by Ry = B% +5'13+(%)w (13) where, hy 1s the fuel heat transfer coefficient, hp, the coolant heat transfer coefficient, & , the wall thickness and k, the thermal conductivity of the wall. Substituting into equation (13), Rp =X 41 . 0017 =z 0.000363 hr-£tZ-CF/Btu T " 5600 ' 18,500 @ (12)(13) T 37 If & film forms on the fuel side of the Inconel tube, the thermal resistance of the system is increased by 0.00025 hr-ftE-OF/Btu. This is equivalent to a 40.8% decrease in heat transfer. Thus, if a film having a high thermal resistance forms in a heat exchanger designed to remove 50 megawatts of heat from the fuel, the heat removal is reduced to 29.6 megawatts and the outlet fluid temperature becomes 1263°F instead of 1100°F. Conclusions From the heat transfer results for sodium hydroxide in nickel tubes and for Flinak in nickel and 316 stainless steel tubes, it can be concluded that both fluids are "ordinary" as far as heat transfer is concerned. In the Flinak-Inconel system an insoluble, high-melting surface deposit (film) results from a reaction at the fluid-metal interface. The thermal resistance of this film is sufficient to cause a marked reduction in heat transfer for Flinak flowing in small diameter Inconel tubes. For systems in which surface deposits do not form, the accepted corre- lations for turbulent forced convection heat transfer (e.g., j = 0.023 NRe~o,2) may be used to predict sodium hydroxide and Flinak heat transfer. If systems in which deposits occur must be used, knowledge of the thermal resistance of the film is required also. _— , 5. APPENDIX A Nomenclature Specific heat of fluid at comstant pressure, Btu/lb. fluid-orF Inside diameter of tube, ft Coefficient of heat transfer in region of fully developed turbulent flow, Btu/hr~ft2-°F; ha, coefficient on coolant side; hp, coefficient on fuel side; hy, coefficient at position x on tube; h' apparent coefficient with film present Thermal conductivity of fluid, Btu/hr-ft-YF; ki, thermal conductivity of film; k;, thermal conductivity of metal Rate of heat transfer, Btu/hr; g, heat gain by fluid; qg , heat loss to environment; ge, electrical heat input Radius, ft; rg, inside wall; r,, outside wall Temperature, oF, te, enviromment; t,, fluid mixed-mean; tm,i, at test section inlet; t, ,, at test section outlet; tm_ave: arithmetic average of tm i and tm,o0; ts, inside tube surface; tg,avesaverage; Tirs outdide tube surface; ty ave, average Mass rate of flow, lbs. fluid/hr Distance down tube from entrance, ft Length to diameter ratio for tube, dimensionless; (x/d)e, at end of thermal entrance region Thickness of film, ft Area of heat transfer surface, ftg; A", inside surface area of film Voltage impressed on test section, volts Mass velocity, lbs. fluid/hr-(ft® of tube cross section) F -39- Current passing through test section, amperes Volume of metal in tube wall, f£t3 Volume heat generation, Btu/hr-ft3 Tube wall thickness, ft Absolute viscosity of fluid, lbs/hr-ft Nusselt modulus, dimensionless, {hd/k) Peclet modulus, dimensionless, Npe'Np.; (cpdG/k) Prandtl modulus, dimensionless, (QP/M/k) Reynolds modulus, dimensionless, (dG/m) Stanton modulus, dimensionless, Ny,/(Npe°Np,.), (h/qu) 2/3 Colburn function, dimensionless, Ngt°Np,. “”' | ~40- 10. 11. 12. 13. 1h. 15. APPENDIX B Bibliography Boelter, L.M.K., Martinelli, R. C., Jonassen, F.; Trans. A.S.M.E., 63, 4bt (1941) Colburn, A. P.; Trans. A.I.Ch.E., 29, 17k (1933) Grele, M. D., Gideon, L.; Nat. Adv. Comm. Aero, RM E52109, Unclassified (1953) Grele, M. D., Gedeon, L.; Nat. Adv. Comm. Aero., Rm E53L18, Secret (1954) Hoffman, H. W.; Oak Ridge National Iaboratory, Physics Report 1370, Unclagsified (1952); Heat Transfer and Fluid Mechanics Institute, p. 83 (1953) Hoffman, H. W.; Oak Ridge National Laboratory, Memorandum CF 53-8-106, Secret (1953) Hoffman, H. W., Claiborne, J.; Unpublished data Hoffman, H. W., Lones, J.; Oak Ridge National Laboratory, Memorandum CF 5k-9-2, Secret (1954) Karman, T. von; Trans. A.S.M.E., 61, 705 (1939) Kirst, W. E., Nagle, W. M., Castner, J. B.; Trans. A.I.Ch.E., 36, 371 (19%0) McAdams, W. H.; "Heat Transmission,” 3rd Edition, p. 219 (195k4) Prandtl, L.; Physik. Zeitschr., 29, 487 (1929) Reynolds, 0.; Proc. Manch. Lit. and Phil. Soc., 1k, 7 (187k4); Collected Papers, 1, 81 (1900), Cambridge Salmon, D. F.; Oak Ridge National Laboratory, Special Report 1716, Secret (1954) Wilson, E. E.; Trans. A.S.M.E., 37, 47 (1915) -41- L. APPENDIX C Physical Properties of Flinak The physical properties of Flinak were obtained from the following sources: RS © Physical Property Charts for Some Reactor Fuels, Coolants, and Miscellaneous Materials - June 1954 - ORNL CF 54-6-188; Powers and Blalock - ORNL CF 53-7-200. The value reported and used in the calculations in this report is, cp = 0.45 * 0.03 Btu/1b-°F, for the temperature range 900°F - 1600°F. Unpublished data of Redmond and Cohen obtained on Brookfield Viscometer; see also Physical Property Charts - ORNL CF 54-6-188. Physical Property Charts - ORNL CF 54-6-188; Cohen and Jones - ORNL-1702.. Physical Property Charts - ORNL CF 54-6-188; Cooper and Claiborne - ORNL CF 52-8-163. ' The value reported and used in the calculations in this report is, k = 2.6 Btu/hr-ft-OF, for the temperature range 1000CF - 1275CF. The viscosity and density are shown as a function of temperature in Figure 16. — ORNL—LR—~DWG 3460 36.0 32.0 28.0 24.0 20.0 16.0 ABSOLUTE VISCOSITY (Ib/hr-ft) 1 1420 140 8.0 4.0 130 120 10 800 1000 1200 1400 TEMPERATURE (°F} 1600 1800 Fig. 16. Variation of Physical Properties of FLINAK with Temperature . —4D — o, DENSITY (Ib/ft?)