MARTIM MARIETTA ENERGY SYSTEWS LEBRARIES 3 4456 0350428 7 CENTRAL RESEARCH LIBRARY & DOCUMENT COLLECTION 2 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 loan. ¥EG IR Pt ORNL -2199 C-84 - Reactors-Special Features of Aircraft Reactors This document consists of 148 pages. Copy f/. of 280 copies. Series A. Contract No. W-7405-eng-26 Reactor Projects Division INVESTIGATION OF FLUID FLOW IN THE ART AND OTHER REFLECTOR -MODERATED REACTOR CORES Muller Bradfute Lynch LD H o DATE ISSUED - », TS L "l'v.\‘r“' " 1Oy T LR T e ’ - T Toad X " o, T (TN UNION CARBIDE CORPORATION for the 3 445 O3 U.S. ATOMIC ENERGY COMMISSION Ouzs 7 -1ii- TABLE OF CONTENTS o Page SUMMARY .ivcvcveeernssnanoonsssssonsososnssaasosnsssas B 1 NOMENCIATURE . ..vvoveesosscessesossssssrsscnsooncocccasssas tseescrsesreas 6 INTRODUCTION . vsvevococssooonosasosssossoovacoccosrssssnossonnsossesessss 10 Statement of Problem ......... S 10 Developmeant of Project and Statement of Purpose ........... et tavens 10 EXPERTMENTAL APPARATUS .....ccc... ceecsiecesaseanenan ceciecosavecarasess 1D Flow Systenl vecevsvesoaee s eeteareesetsarrerens cesescenens ceveerannne 15 The Core Models ...cce0ce sevses sessassatssasns settsesssccarssosannns 15 Core Entrance Systems ......... ceeessessssssesaanaans ceccane ceesecannn 21 MATHEMATICAL ANALYSES ....0cececeacranssanaas cecesenan cececracnnnsiases 27 Mean Velocities and Reynolds Numbers in "2l-inch" ART Core ......... 27 Pressure Distributions in ART COTre€ ..cieseecroconsssessssossonsasssos 29 Radial Pressure Differences ...e.ceeeesecens eestectesres st on e cecane 33 Core Desigr. Using the Nikuradse-Buri Parameter I' for rlow in Divergent Channels ...cc.ceceevreccosrccssccnes recesrersesssssassa 3 Pressure Distribution in ART Heads .eeicevereseccccccnnanes ceerecaae 37 EXPERTMENTAL ERESULTS ........ ceesescesaas tcesecncene cresestesanasena cease Ll Axial Flow Through Cores ...... Cesesneanaas cetececsscevesncocecnness Bl Rotational Flow Through COres ........ ceeeenreanan cecoecrecosenaan oo Lo Flow Through 10/44-Scale ART "21-inch" Core - Twin Tangential Entrance Header - One-Pump Operation ..... ceceesesascasrssaarao e 55 Flow Through Cores with Turbulence Promotion ....c.ecececesvvoceocass 55 —iv— ! R ey AT Page /;$Ve10city Fluctuations in the Full-Scale ''21-inch" ART Core Model .. 70 Fluid Velocities in the Full-Scale ART '"2l1-inch" Core ............. 82 Pressure Distributions in '"21-inch" ART Core .......... Cer e 82 DISCUSSION - Literature SULVEY .....civertersoronsntosansonsnsonanasass 89 Flow in Convergent and Divergent Channels ............ Ce e R 89 Flow in Curved Channels .......... e Che et 95 Flow Through SCTeens .......iceeivsnonssacsooonsnsnossscnsnsnnans - 98 Screens in Diffusers .................. C et e st et 101 DISCUSSION - Discussion of Results .............. C e 103 Axial Flow Through ART Core Models ........... Chee e Ce e 103 Axial Flow in Two Constant-Gap Core Models .......... Cehe e e e 104 Rotational Flow Through ART Cores ........... N 104 One-Pump Operation - ART Inlet Header ........ c e et 105 Flow Through Cores with Turbulence Promotion ...... Cee e ... 105 Flow Through Cores Packed with Screemns.............. .. iiueun, .... 109 Velocity Fluctuation Studies .............. L ec e s e c e eaan e ceees 113 Average Fluid Velocities in the Full-Scale "2l-inch" ART Core ..... 114 Pressure Distributions in the "21-inch" ART Core .................. 116 Pressure Distributions in ART Heads .........c. iiiiiiiiiinnonnnnns 116 CONCLUSIONS .......... Gt s e et e s e e e ceaer e e rera e 118 APPENDIX A ........... e f e e e ees e s T 122 Velocity Measurement and Flow Visualization Technique ............. 122 APPENDIX B .....c0vvvuns C et bt 133 Flow System Components ........ ettt eetaaas Cearereeersenana Cehaae 133 REFERENCES ....... C it s s e C ettt ee e . 136 INVESTIGATION OF FLUID FLOW IN THE ART AND OTHER REFLECTOR-MODERATED REACTOR CORES SUMMARY The turbulent flow of liquid through ART reflector-moderated reactor cores of annular cross section has been studied because of the need to ob- tain a steady flow exclusive of the normal unsteadiness of turbulent flow in ducts. This steadiness is important because when a large smount of heat (100,000,000 Btu/hr-ft3) 1s being generated in the volume of liquid by nu- clear fission at high temperatures (lEOOOF - 16OOOF), low-frequency temper- ature fluctuations due to unsteady, irregular flow of the liquid can result in hot spots, thermal cycling, and fatigue of the containing materials. To facilitate the experimental work, studies were made of water flow- ing through quarter-scale trafisparent plastic models of the cores so that flow visualization techniques could be used in the work. An extensive series of tests of many core systems were made over a range of Reynolds moduli based on core midplane dimensions and axial flow rate up to ~,90,000. The corresponding Reynolds modulus for the "2l-inch" ART core was ~. 95,000, using the properties of fluoride mixture No. 30 at 1h250F. Also, analyses were made .of velocity and static pressure data ob- teined oa a full-size "21-inch" ART core model with several different en- trance conditions by Whitman, Stelzman, and Furgerson of ARED. The following conclusions were reached as a result of this series of experiments: - . - . Ir e nmmfimmm‘b I fimx’n*".{}:qg‘.,xg . (a) (b) (c) Axial flow through the cores with sufficiently large cross-sectional area expansion rates considered here will always be accompanied by a separation of the forward flow from a point on the outer wall near the inlet and viclently unsteady reverse flow in the "separated" re- gion. This separation at the outer wall is due to the large adverse rressure gradient resulting from the large area expansion rate, the lower fluid shear stress at the outer wall than at the inner wall {common to flow through annular ducts), and the curvature of the channel. The flow into the entrance of the core models was always uniform radially and circumferentially, having passed through a calm- ing length of 40 diameters of straight pipe into an annular "nozzle" which was mounted on the core entrance. Rotational, or spiral, fiow through the same cores is always accom- panied by separation of the forward flow from a point on the inner well near the inlet and unsteady reverse flow in the separated re- gion with a velocity component in the direction of the rotation. The separation in this case is caused by the same adverse pressure gradient and the lower fluid shear stress at the inner wall relative to the outer wall (a phenomenon of flow in curved channels). Agsain, the flow at the core entrance was uniform radially and circumferen- tlally, having passed through 40 diameters of straight pipe into an annuler "nozzle" which preceded the turning-vene section which pro- duced the rotational flow. In rotational flow, the turbulent interchange of momentum and the eddy S -3- LI conductivity are diminished at the inner wall due to the centrifugal force field set up by the fluid motion; and, therefore, increased tem- peratures due to volume heat generation can be expected at the inner wall over the temperatures obtained in straight flow at the same rate in an equivalent channel. (d) If the spiral velocities are very high, frictional forces become of such megnitude that a decay of the spiral velocities becomes notice- able and a backflow due to spiral vortex decay will begin. fhis was noticed in the high spiral velocity case studied by Whitman, Stelzman, and Furgerson. Backflow occurred on the inner well at the exit.plane. (e) Turbulence promotion within the cores has been shown to overcome the effects of adverse pressure gradients, but the promoters must be such that they do not introduce unsteady flow; i.e., the scale of the pro- moted turbulence must be very small. Woven-wire screens and perforated- plate screens of low solidity and small wire diameter (or web thickness) packed into the divergent part of the core flow channels are of such nature. Turbulence promoters such as vortex generators or large ob- structions placed in the core entrance cause unsteady flow although they seem to eliminate the separation and the associated backflows. (f) Straight annular cores and cores of sufficiently low expansion rates can be constructed to give a steady, unseparated flow with axial flow. Calculations show tpat the meximum midplane-to-inlet area ratio that can be achieved with axial flow through & bare core is 1.33:1 within an 18-in. length of a 21-in.-0.D. core with the same midplane area G b £ {(g) (h) as the "21-inch"” ART core. Calculations have shown that a pressure unbaslance (L lb/in.e) due to momentum transfer exists in the header which is of the same order as the friction losses in the core., This pressure discrepancy is in the form of a rise in pressure as the fluid traverses the length of the header from the inlet duct. The unbalance, plus any unsteadiness trans- mitted to the flow by the fuel pumps, will also creste peripheral flow asymietries and unsteady core flow, The calculations also point up two considerations. One is that the header pressure unbalances are primarily due to the average fluid velocity level in the headers considered. Thus, header fluid veloc- ities should be kept as low as practicable. The core pressure 1loss should also be as large as practical to keep the relative importance of the header unbalances small. Since screen packing in the core has been shown to eliminate the un- steady flow, which is an inherent characteristic of the core shape, it 1s proposed that this system be used in the core. An additional sdvan- tage accrues from the use of the screens in an increase in core pres- sure loss and the related velocity profile flattening. The peripheral asymmetry due to pressure unbalances in the header will then be much less than exists without the screens. It is felt that perforated-plate screens with the same relative pressure loss and mesh size as the wire screens tested would be more advantageous from the structural and fab- rication standpoint than the wire screens. Foo i M N -5- The heat-transfer characteristics of the screens are in the proc- ess of being investigated to determine whether any problems exist in this regard. (1) A header system has been designed which may afford smaller pressure un- balances than the present header system and, being used in conjunction with the screen-packed core, may allow single-pump operation without large peripheral flow asymmetries. Further experimental work is in process to establish the validity of this conclusion and conclusions (g) and (h). = = Re Aol -6- fi;-@ L, NOMENCLATURE constant, f’t2 core cross-sectional area perpendicular to core axis, ft2 header cross-sectional area perpendicular to flow direction, ft2 constant, ft constant, fte/sec constant, £72 2 constant, ft typical dimension of screens; wire diameter, bar width, effective web thickness of perforated-plate screen (d = M - 0.95 D), ete., ft hole diameter in perforated-plate screen, ft hydraulic diemeter, ft 7 eddy diffusivity, equal to -E , Tt°/sec coefficient of friction in Fanning equation, dimensionless gravitational constant, ft/sec2 head, f't head at position 1, ft head at position 2, ft mixing length according to Prandtl, ft flow path length in core, ft axial length of core, ft spiral flow path length in core, ft mesh size of screen, ft Reynolds modulus, dimensionless =% NRe,sc NRe,avg Reynolds modulus based on d, typical dimension of screen, dimensionless axial flow Reynolds modulus based on average core dimensions and mass flow rate in core, dimensionless axial flow Reynolds modulus based on core midplane or equator dimen- sions and mass flow rate in core, dimensionless pressure, lb/ft2 pressure at inner core wall, lb/ft2 pressure at outer core wall, lb/f‘l:2 pressure at position 1, including pressure due to elevation, 1b/ft2 pressuré at position 2, including pressure due 1o elevation, lb/ft2 pressure at O-deg plane in header, lb/ft2 (see Figure 20) pressure at 90-deg plane in header, lb/ft2 (see Figure 20) pressure at 130-deg plane in header, 1b/ft2 (see Figure 20) pressure difference, lb/ft2 axisl-flow pressure difference across core, lb/ft2 spiral~-flow pressure difference across core, lb/ft2 volume flow rate, ftg/sec radius, ft radius of inner weall of core, ft radius of centerline of core flow channel, ft radius of outer wall of core, ft ratio of spiral flow path length to core length, dimensionless a length along spiral flow path, ft half-width of core flow channel perpendicular to channel centerline, ft root-mean-square of fluctuating velocity component in direction per- pendicular to plane of channel walls in two-dimensionasl flow, ft/sec 8. 2 axial component of velocity in core, ft/sec average of above, ft/sec fluid velocity in header, ft/sec fluid velocity at inlet of header, ft/sec mesn velocity in channel at any cross section, ft/sec maximum cf above, ft/sec 2 t,ave * Ya,ave ’ ft/sec average spiral velocity in core, equal to flJ-v peripheral, rotational, or tangential velocity component, ft/sec average of tangential velocity component in core, ft/sec velocity at position 1, ft/sec velocity at position 2, ft/sec root-mean-square of fluctuating velocity component in direction of flow, ft/sec mass flow rate in header, lb/sec mess flow rate at inlet of headér, 1b/sec distance parallel to channel wall in direction of flow, ft distance from core inlet along core axis, ft distance from header O-degree plane along mean length of header, ft (see Figure 20) distance downstream from screen plane, ft mean length of header, ft differential element of length in direction of flow, ft distance from channel wall perpendicular to plane of channel wall, ft differential element of length perpendicular to plane of channel wall, ft distance from core midplane parallel to core axis, ft _.9..: angle between spiral-flow component and plane perpendicular to core axis, deg ‘ density, lb/f'b3 boundary layer thickness, ft turbulence factor, equal to pl2 %% ’ lb--sec/ft2 o momentum thickness of boundary layer, equal to.f (1 - —) X dy, ft 0 Vmax mex coefficient of viscosity, lb-sec/ft2 . . K 2 kinematic viscosity, equal to 0’ ft /sec mass density, lb-—sece/-ft)+ shear stress, lb/ft2 — -10- fié;:i v INTRODUCTION *n Statement of Problem The design of circulating-fuel nuclear reactor systems which must operate at high power densities (~-100,000,000 th/hr‘ft3) and at temperatures near to the limits of endurance of their containers (lEOOoF - l600°F) places special emphasis on the detailed knowledge of the behavior of the circulating-fuel tem- peratures with respect to position and time. According to analyses on forced-convection heat transfer with volume heat sources within the liquid,l’ 2, 3 the temperature distribution depends upon the velocity distribution, the heat-generation rate, the heat-removal rate at the wall, and the molecular and turbulent heat transport properties of the liquid. If the velocities and turbulent heat-transfer properties are unstable with time and asymmetric with respect to the channel center and core circumference, then fluctuating temperature distributions and hot spots will occur due to the asym- metrical heat removal from the core. The magnitude of the temperature oscillé- tions and hot spots may or may not exceed the endurance limits of the containing materials. The problem took two parts: the first, to find some way of obtaining sta- ble and peripherally symmefrical flow in the ART cores proposed by A. P, Fraas and W, T. Furgerson of the Osk Ridge National Leboratory Aircraft Reactor Engi- neering Division; and second, to ihvestigate other designs for reflector- fl moderated reactor cores in an effort to find some configurations which would not have unstable flow, Development of Project and Statement of Purpose In 1952, R. E. BalllF observed the flow of air through a configuration wni called the "Fireball" reactor core which had an annular cross section prepen- dicular to the flow. The center-to-inlet (or exit) area ratio was 5.4:1 while the total length of the varying cross section was 24.6 in., A section parallel to the core axis is seen in Figure 1. Ball found that the forward flow separated from the outer shell at approx- imately 5.3 in. from the inlet and that the separation increased in sevefity with distance along the flow path. By filling in the region of separation using plasticene, Ball was able to provide a flow contour which reduced the amount of reverse flow. Since the velocities, as measured with a pitot tube, were still low near the walls, vanes were added in the inlet which, after some exfierimen- tation, produced a fairly flat velocity distribution. However, no observations were made to determine fluctuations in the velocity with time. Ball states that some pecularities in forming the Plexiglas outer shell of the core resulted in a greater rate of divergence than anticipated. However, since the originally designed rate of divergence was so much greater than the limiting maximum value of 8 deg included angle for a conical diffuser in which no separation takes place, separation would have occurred even without the vari- ations introduced in fabrication. Filling in the flow passage reduced the rate of divergence of the channel. .Evidently, Ball's results were deemed inconclusive either because .of the inaccuracies in fabricating the Plexiglas core shells or because of the reduced core diameter resulting from the "filling in", since another investigation was begun by Stumpf of ARED.26 During the second quarter of 1954, the experimental > and H. F. Poppendiek. study reported here was initiated by J. 0. Bradfute —§2- ORNL-LR-DWG 2244 9 10—in. DIA. 6-in, DIA, 9-in. DIA. _ 6—in. DIA. ) N N N | ORIGINAL CONTOUR g . N r;_:) g It \ T 9-in. DIA. __[ . 3/ £ ‘ 1794 ~in. DIA; g -1 9 \ 15.7—-in. DIA. © \ o X\ £ Q Ty \ Fig. 1. Cross-Section of Mode! of "Fireball" Reactor Core (Reference 4) r Lyt m A B Cur, }% E,:a(«{ capall s N 36 in. 18 in, Fig. 3. Cross—Section of 21-in. ART Core Flow Channel Bradfute studied the flow of water through a quarter-scale model of a core with an equatorial outer diameter of 18 in. This core was assembled using stainless steel for the inner shell and Plexiglas for the outer shell, both shells being carefully machined to the correct contours. The shape of this "18-inch" core differed from that studied by Ball and is shown in Figure 2. The full-scale core had a midplane-to-inlet area ratio of 3.8:1 and an over-all length of 36 in., The reduction in the area expansion ratio was effected by increasing the outer shell diameter at the inlet from 8 in. to 10 in. This change, coupled with an increase in core length, resulted in a reduction in the channel divergence angle. It was later found32 that the "18-inch" core required a larger UFA con- centration (6.5 to 9 mole %) than was originally calculated (~.4 mole %). The new calculations showed that a 21-in.-dia core would be required to obtain the 4 mole % concentration. The core shape was proposed by W. T. Furgerson. This design had a midplane-to-inlet area ratio of h.36;l in the same length as the "18-inch" core. This core is shown in Figure 3. Early in 1955 work was in- itiated on a 10/44-scale model of this "21-inch" core. Parallel to this investigation, full-size "21-inch" core mock-up studies were carried out by G. D. Whitman, W. J. Stelzman, and W. T. Furgerson using the header and pump system designed by Furgerson et al. “ Studies of circulating-fuel reactor core hydrodynamics have also been made at Pratt and Whitney Aircraft in conjunction with their reactor research program_33-uh Mention of these studies is made so that a complete reference to all published reports known to the authors on this subject is at hand. af -— 2 Due to the lack of an adequate analytical technique for establishing ey velocity distributions in geometries as complex as those of the ART cores, re- course was made to the experimental solution of the problem. The techniques described were developed to provide a means for rapid qualitative examination of the velocity distributions in the test core geometries. Since it was felt that the phenomena of separation, backflow, and large- scale unsteady flow observed in the "18-inch"” and "21-inch" cores were un- desirable for a high-temperature, high-power-density core, much emphasis was placed on the effort to find reflector-moderated reactor core systems which possessed steady flow. Most of the velocity distributions shown were sketched from visual obser- vations of the deformation of a glowing band of excited phosphor particles as they moved with the water through the core models. Measurements of the veloc- ity distributions for the case of axial flow through the "18-inch™ ART core model were made by the stroboscopic photography of tobacco seeds suspended in the flowing water. Analyses of the data obtained by the group studying flow in the full- size model of the "21-inch" core with several different entrance systems are also presented. — 13- EXPERTMENTAL APPARATUS Flow System The experimental flow system consisted of a recirculating loop contain- ing fluid reservoirs, a centrifugal pump, flow and temperature measurement and control devices, and a test section containing the core models and associ- ated entrance and exit regions. A schematic drawing of the system is shown in Pigure 4. PFurther details regarding the system components are presented in Appendix B. Water was used as the working fluid. Both stroboscopic particle photo- graphy and phosphorescent particle visualization techniques were used to establish the local velocities and flow characteristics of the core regions. These methods are discussed in greater detail in Appendix A. The over-all friction losses in the cores were determined using a 30-in., U-tube mercury- manometer, Inlet and exit fluid temperatures were measured with mercury- glass thermometers. The core models and entrance systems studied are de- scribed in the following paragraphs. The Core Models The core models studied are tabulated below: a. A 1/4-Scale Model of the "18-inch" ART Core. The outer shell was machined from a solid Plexiglas piece and the island was machined from stain- less steel. Figure 5 shows a cross-sectional view of this test section with a table of typical dimensions. The dimensions of the inner and outer walls of the full-size core are obtained from the cylinders of revolution described SURGE CHAMBER -16- ORIFICE UNCLASSIFIED ORNL-LR-DWG 7170 STIRRER TANK MERCURY MANOMETER T CENTRIFUGAL PUMP Fig. 4. Schematic Drawing of Experimental Flow System. 1] TEST SECTION CONTROL VALVES *TO DRAIN Haw, PLEXIGLAS SHELL DISTANCE FROM MIDPLANE (in.)) INNER DIAMETER (in.) OUTER DIAMETER (in.} oo Fig. 5. | TR —_ 7 ORNL—LR—-DEZZ44 I I rY k=g r A nc-,;_:,—-\: 1 e STAINLESS STEEL ISLAND ][ MIDPLANE T Y [ ) LA Zk 7/ % ' [/l=—TEST SECTION SUPPORT [ ] 2] I Z g / ! 0N G 1 N H : s 1 N N 7 N LA ] ;\ | TN N | /l 1 | 0 05 10 15 20 25 30 35 40 45 2.250 2.162 1.940 1.688 1.523 1.500 1.500 1.500 4.500 {.50C 4.500 4.42) 4.198 3.870 3493 3133 2.851 2.692 2.574 2.50C Cross Section of Quaorter-Scale 48-in. ART Core Model. _— 18- by the equations, r = 1.8230 cos %% + 7.1770 for 0 | Y A%]/ Fig.10. Cross Section of 24-in. ART Core Model Using Screen Packing in Expanding Entrance Half of Channel. ORNL-LR-DWG 22442 ENTRANCE SECTION TEST SECTION SUPPORT TEST SECTION UPPER PIPE FROM PUMP Y Z —=— COLLAR — /\\\\\.\\ \\\\\\\\\\\\\L BRASS SLEEVE RETURN PIPE Test Stand Assembly. Fig. 11, L ML D PO EE Fig. 14. The Quorter-Scaole Model of the 18-in. Core with the Prott ond Whitney Vortex Generator gt the Inlet Fig. 12. Turning-Vone Section angential-Intet Header Model Fig.43. Twin T ...gz_ _—— -27- MATHEMATTCAL ANALYSES Mean Velocities and Reynolds Numbers in "21-inch" ART Core Mean axial and rotational components of velcocity were obtained for three entrance conditions for the "21-inch" ART core from velocity data tsken ifi the full-size ART core model.6 The entrance conditions were: (1) a high velééity rotational flow induced by two 2.1 by 2.1-in. tangential entrance ducts directed perpendicular to the core axis, (2) a lower velocity rotational flow induced by two similarly arranged 4.38 by 3.45-in. ducts;, and (3) a still lower velocity rotational flow induced by adding turning vanes to the second case. Axial components were obtained by dividing the volume flow rate by the core cross-sectional areas. Dividing the volume flow rate by the entrance duct areas for the first two cases gave the meen rotational component at the entrance. By assuming conservation of angular momentum, the mean rotational components were computed as a function of axial length. For the third case (turning vanes), the rotational component was obtained from velocity data6 measured at the core midplane. This data was first adjusted to account for the property differences between fuel No. 30 and water; and then, using angu- lar momentum conservation, the radial component was calculated as a function of axial position. The velocities obtained from these analyses are plotted as a function of axial length in Figure 15. ‘The velocity results were combined with the core dimensions and the phys- ical properties of fuel No. 30 at 1425°F (Table 1) to obtain the resultant vector Reynolds moduli. ORNL—LR—D!G 22444 °0 | T T — ~—— AXIAL COMPONENT — ROTATIONAL COMPONENT - 40 \\ 4/ w > \ HIGH VELOCITY ROTATIONAL FLOW ’/ — 30 e, \ / S - L ~ 20 =2 0, there is obtained = csC Q. KI& axial length z. Thus, 18 18 L = Jr csc @ dz = f(z)dz (4) -18 -18 The functior f(z) is plotted in Figure 17 for the two cases, The integration was performed graphically. L The ratio of spiral flow path length to axial length (ii = 19 thus a * Calculations show that the length of the axial flow path is only 2% longer (36.78 in.) than the axisl length of the core, so that the equation is a good approximation. RC TINUL LD VIUUULUD Y. ¥ T 5 F (xqéo ) ORNL—LR—DWG 22440 pd N 15 /r \\ 14 e N HIGH VELOCITY ROTATIONAL FLOW 2 // N\ 11 \ 10 9 8 ? Ju— 6 S , — T MEDIUM VELOCITY ROTATIONAL FLOW 3 2 — LOW VELOCITY ROTATIONAL FII_OW (IGS—-Z TURNING \/ANESL____ I * — —— AXIAL FLOW 0 | | I 18 16 14 2 10 8 6 4 2 0 2 4 6 10 12 14 16 18 DISTANCE FROM CORE MIDPLANE (in.) Fig. 16. Prediction of Resultant Vector Reynolds Moduli in 24-inch ART Core for Flow of Fluoride Composition No. 30 at 1200 gal/min. 18 16 D COSECANT, a, R 10 o wti Ry b COSECANT, a, W 30 25 n O ANGLE — DEGREES O ORNL-LR-DWG. 12749 S // 4q 6 8 10 12 DISTANCE FROM EQUATOR Fig. 17. The Cosecant a as a Function of z. 14 16 - le_ —_— -32- determined was used to calculate the ratio of rotational flow pressure loss to axial flow pressure loss according to the equation, A L v 1.75 v 1.75 PS _ 5 S,avg - R S!avg (5) ap, L 1.75 va,avg . N a a 8a,avg Since the average core velocity is directly proportional to the flow fifith length, it follows, assuming equal volumetric flow rates that Vs,av = R 7 a,avg and hence, - 2.75 Ap, = op - R . {6) The axial flow pressure loss was calculated for an annulus of the follow- ing average ART dimensions: r, = 8.427 in. r. = 4,653 in. La = 34.3 in, and for va,avg = 2.5 ft/sec then NRe,avg = 116,500 T = 0.018 from which L v 2 op. = £ == —28Y 5 011 1b/in.? (7) a Dh 2gc The ratio R for case 1 was found to be 11.7 and for case 2, 3.6. The calculated spiral-flow pressure drops are then 9.7 lb/in.2 for case 1 and 0.41 lb/in.2 for case 2. The core entrance losses were determined using the Darcy-Weisbach equa- tion for the friction losses in the entrance ducts and the pressure data from case 2 for the losses incurred as the fluid turns into the core. Since there were no similar data for case 1, it was assumed that the turning losses were proportional to the square of the velocity coming from the entrance ducts. Table 2 gives the resulting core entrance losses. TABLE 2 Core Entrance Pressure lLosses Flow Rate - 1200 gpm Case No, 1 Case No. 2 Friction loss in ducts between pumps and core 3.0 lb/in.2 0.2 lb/in.2 * Turning loss - 36.6 1b/in.? 3.1 1b/in.° Radial Pressure Differences To obtain some idea as to the effect of the rotational component of veloc- ity on the radiasl pressure distribution, it was assumed: (a) that a free vortex-type rotationsl velocity component exists; c i.e., Ve =T and, (b) that the total contribution to the radial presgure difference is the v rotational velocity component; thus, %E =P *%- . Integrating between the inner and outer walls, there results 1 1 1 p s B (- ) (8) r r % L % i 0 ¥ Measured pressure loss from reference 6, o -3h- o In Figure 18 the predicted radial pressure difference between the outer and inner walls, D, - Py is shown for both case 1 and case 2. Core Design Using the Nikuradse-Buri Parameter I for Flow in Divergent Channels 7 Buri's corrections’ of Nikuradse's results8 for turbulent flow in diver- gent channels show that the wall shear stress drops to zero, indicating bound- ary layer separation, when the dimensionless parameter dv e max 0.25 v dx NRe,e (9) I' = v max has a minimum value of - 0,060, where max NRe,@ TV vfi t Taking I' = - 0.057 and using the ratios - = 0.735 and 5= 6.80 as found max by Nikuradse, eguation (9) can be rewritten to give L dym Vfit 0.25 P' == "'F— a"x"'* —\7— - - 0-57“‘ . (10) m This new parameter, I'', can then be used to design a convergent-divergent core having no flow separation at the walls. From continuity requirements, A, v, =Q=a constant (11) for all flow cross sections. Then equation (10) becomes Y 60 50 40 W o p,—p b/ %) 20 10 ORNL—ILR—DWG 22439 DISTANCE FROM INLET (in.) \\CASE NO. 4// e ———" CASE NO. 2 ] | s sl I I 0 4 8 12 16 20 24 28 32 Fig. 18. Predicted Pressure Difference Between Outer and Inner Core Walls for Flow of Fluoride Composition No. 30 at 1200 gal/min. 34 _ge_ - -36- or on substituting, A =rrt, x ¥x C m dr 0.25 m 0.57h (}_1_1_2) 157t - \a dx . | (12) m Integrating equation (12) under the conditions that (1) the core midplane area is same as that of the ART (1.775 fta), (2) the outer shell diemeter is 1.833 ft at the midplane, (3) the channel half-thickness, t, is 0.196 ft, . (4) the flow rate, Q, is 2.70 ft%/sec, (5) the fluid is the fluoride mixture No. 30 at 1425°F; hence, v = 1.35 x 10™° £t2/sec; and, (6) r, = 0.721 ft at x, = 1.5 ft, r X m c drm ] 0.514 (_);"__TT_Y_).25 dx 1.25 t Q c rm 0.721 1.5 ) .25 -4 (r '8 . 1.085) = 5.5(%) (x, - 1.5) or 4 ) 'm = [- 0.2605 (x - 1.5) + l+.3!+o_| ' (13) Table 3 shows the results of the evaluation of equation (13). - -37- TABLE 3 Variation of rm as a Function of xc B w i@...ffl 5 0.512 0.3 0.546 0.6 0.585 0.9 0.627 1.2 0.672 1.5 0.721 Figure 19 presents the resulting core geometry as constructed graph- ically from the information of Table 3. Near the core midplane, the walls are rounded off to give a smooth transition to the symmetrical exit half. The re- sulting midplane outer diameter is 1.75 ft (21 in.) and the midplane-to-inlet area ratio is 1.33:1. Without rounding-off at the midplane, the ratio would be 1.41:1. In comparison, the "18-inch" and "21-inch" ART cores have area ratios of 3.80:1 and 4.36:1, respectively, for the same length as the 1.33:1 area ratio core. Pressure Distribution in ART Heads Pressure_unbalances in entrance regions such as the ART core inlet head are caused by friction losses and momentum changes. For the systems under study, friction losses were found to be small as compared to the pressure changes due to momentum transfer. The pressure variations due to momentum v I T MY I AINNIITT Ar AT AN CINITnANGVLE Vi) ORNL—LR—DWG 22438 0.2 — 04 (— | o o o | o 1Al 1T g - l r / | / g - 2 ] s ‘ ~ / ~ [ / = 3ftgp /\ __ f 2.608 fTN / \\ : CORE MIDPLANE - 7z \ V A _ I ] 1.0 0.8 0.6 0.4 0.2 0 0.2 0.4 0.6 0.8 1.0 RADIUS FROM CORE AXIS (ft) Fig.19. Cross Section of Inlet Half of Annular Flow Channel of Core Constructed from I'-Function. ~ ™ o o -39- changes have two sources. One 1s the Bernoulli effect, in which a velocity change in the fluid is accompanied by a pressure change. The second is the momentum transfer that occurs as the fluid leaves the header and turns into the core. While data of other investigator39 indicate that the momentum transfer is less than perfect, a 100% effectiveness was assumed for simplic- ity and conservatism in computing the pressure changes. Then, the equation, _ d(wvh) Mp8e is obtained, assuming that the flow is withdrawn uniformly per unit head area. dp = (1k4) Integrating equation (1L4), woo-tp a2 (L) 15) &c Ah 78, Ah Ah With uniform flow withdrawal per unit head area, the flow through the header et any position, Xy is described by ) v (16) A good approximation is that the cross-sectionsl header area varies linearly with distance from inlet; i.e., Ah=a—bxh o (17) Substituting equations (16) and (17) into equation (15) and rearranging, X 2 2 2 wi Eaxh - 2aX - bxh + bX Ap = - . (18) ’ 7ch2 (a - bx)> " 0 Using the method of partial fractions, -40- X X 2 wi 1 - bdxh a2 o - bdxh Ap:-— 5 -—2- “a-'-—:—-fi';h- ']5 + bX - 28X ""rf"‘——'""——3- (19) rg X b (a - bxh) 0 0 This becomes 2 Vi a - bX 1 1 e [y (o) o (2 - 1) - 78X 2a 2(a - bX) where 1 C — 1 b2 2 8. 2 caX C - - - X + - ° 2 22 b Evaluating equation (20) for the header shown in Figure 20(a) with X = 0.742 ft a = 0.152 ft2 b = 0.0539 % w,® = 71,500 1b°/sec” y = 200 1b/ft3 The pressure difference, Pl30° = Pgo » is found to be 3.6 lbs/in.2 Similarly, for the header shown in Figure 20(b), with X =0.98 ft a <= 0.072 ft° b = 0.049 ft w,% = 71,500 1b°/sec” y =200 1b/ft3 ‘fififii7¥ ? 'lllll'.' _41_ ORNL-LR-DWG 22437 Wt CORE ENTRANCE YA\ N\ ART "F" PUMP VOLUTE 7 (TYPICAL) + e 10Y2 in.—smime10Y2 in.—= ) DO\ DN 3, . N 4 /8 n. Q NN N N SECTION A=A 20(ag) CORE ENTRANCE A\ i o ART "F" PUMP VOLUTE (TYPICAL) 4 < 10% in. 10Y5 in—= SN SOOI 43/gin. ! N N | N % N N SECTION A—A 20 (b) CORE ENTRANCE ’L N A S AN L YN WG N N N ART "F" PUMP VOLUTE ‘ ~ 10% in, 10Y5 in. (TYPICAL) N N DO 43/gin., ) ] § N J TR Y N ~N ~ SECTION A—-A 20({(c) Fig. 20. Core Entrance Header Configurations. o o ho- 2 P1300 - Pgo = 20 1b/in.<, The header shown in Figure 20(c) is of constant cross-sectional flow area. Thus, equation (14) becomes 2 2 dp = - a(v, <) . (21) &, h Integrating equation (21), there is obtained o 2 V. _ 2 _ hi Ap = 2 v, dvy = 2 . (22) C C Vhi Then for Vpg = 514 ft/sec y = 200 lb/ft3, 2 Pgoo = Poo = 1.1% 1b/in.“, The pressure variation as a function of circumferential position for the three headers considered is shown in Figure 21. 20 16 12 .2) P =Dgo (Ib/in Fluoride Mixture No. 30 Flow at ART Design Flow Rate. < !A".-'n‘?'g -,*.h,‘-w‘s o e _43._ ORNL-LR-DWG 22436 / 4 / 18(a) / 18(c) 20 40 ANGULAR DISTANCE FROM O-deg LINE INDICATED IN FIGURE 18 (deq) 60 80 100 120 Fig. 24. Header Static Pressure as a Function of Circumferential Position for 140 -— - EXPERIMENTAL RESULTS Axisl Flow Through Cores (2) (o) (c) The qualitative velocity distributions for axial flow through a qQuarter- scale model of the "18-inch" ART corelO are shown in Figure 22 along a v%?tical section through the flow channel for a midplane Reynolds modulus of 3,000. These distributions represent the visual impressions of veloc~ ity profiles obtained using the phosphorescent particle flow visualization technique described in Appendix A. Although there were some differences of opinion among three independent observers regarding a few details such as the exact shape of the profiles, there was unanimous agreement concern- ing the gross features of the flow. Velocity measurements obtainedll by the stroboscopic photography of par- ticles suspended in the water flowing through the "18-inch" core model at a higher flow rate (NRe = 10,000) verified the qualitative observa- tions made earlier. Figure 23 shows these profiles. The stroboscopic particle photogrephy technique is described in Appendix A. The qualitative velocity distributions in a 10/Lh-scale model of the "2]-inch" ART core > are shown in Figure 24. These profiles were ob- tained visually as in case (a) for a midplane N, = 2,800. In addition, the gross features (flow separastion, etc.) were obtained at a midplane NRe = 18,500 by visually observing the particle motion under a strong light. This method is also described in Appendix A. Figure 25 shows the gross flow features observed under these conditions. ~45- LR Py ek ORNL—LR—DWG 22432 NEGATIVE PROFILES DIFFICULT TO OBSERVE DUE TO LARGE SCALE TURBULENCE VELOCITY IS QUITE LOW AND MAY BE NEGATIVE INDICATING SEPARATION 1 LIMIT OF MAXIMUM PENETRATION OF TURBULENT EDDIES IS OBSCURE DUE TO ONSET OF INTENSE TURBU- LENCE Fig. 22. Qualitative Estimate of Axial~Flow Velocity Profiles in the Quarter-Scale Model of 18~Inch ART Core at N ., = 3,000, Re, mid FoirE T, ORNL -LR-DWG 7197 INLET REYNOLDS NUMBER 10,000 FLOW RATE 1140 cm° /sec VELOCITY SCALE Roomfse, LINEAR SCALE |—3 Tt Fig. 23. Velocity Distribution in the Quarter-Scale Model of the 18-Inch ART Core. -47- ORNL—LR—DWG 22434 LARGE REVERSE FLOW FLUCTUATING IN MAGNITUDE LARGE TURBULENT EDDYING TAKES PLACE DOTTED PROFILES INDICATE THAT THEY ARE LESS FREQUENTLY SEEN DOTTED LINES INDICATE EXTENT OF SEPARATION REGION Fig. 24, Qualitative Observations of Velocity Distribution for Axial Flow Through the "21-Inch'' ART Core at NRe mid = 2,800, N - 48- ORNL—LR—DWG 22435 DOTTED LINES INDICATE EXTENT OF SEPARATION REGION. ARROWS INDICATE GENERAL DIRECTION OF FLOW. FLUCTUATING, TURBULENT EDDIES LARGE AREA OF BACKFLOW. Fig. 25. Gross Flow Features of Axial Flow Through the 21~Inch ART Core at Re, mid =18,500. (d) Visually observed qualitative velocity distributions in a constant-gep- width core of midplene-to-inlet area ratio of l.hh3:ll3 are shown in Figure 26. The gross flow features were obtained by observing the par- ticle motion in the water up to NRe mid = 30,000. The variation of the 2 separation point with Np ., is also tabulated in Figure 26. » (e) Quelitative velocity distributions are shown for a 2.133:1 midplane-to- 13 inlet area ratio core in Figure 27. The velocity distributions were obtained visually using the phosphorescent particle technique at a HRe, nig = 3,000, and the gross flow features were observed at NRe, mid up to 20,000, Rotational Flow Through Cores (a) Qualitatively observed axisl velocity components of the rotational flow through the "18-inch” core msod.ell}+ are shown in Figure 28. The rota- tional component was added to the flow by a set of turning vanes de- signed to give a velocity of constant scalar magnitude at an angle of k5 deg from a horizontal plane at the core inlet. (b) Rotational flow was also observed in the 10/44-scale "21-inch" core 15 model. The qualitative axial velocity component distributions are shown in Figure 29. Approximate flow angles obtained with the phospho- rescent particle technique are also given. The inlet turning vanes (designed by é. F. Wislicenus), which add the rotational component, were shaped such that the trailing edges of the vanes gave a greater rotational component at the channel center than near the walls. The average core in- let flow angle was approximately 45 deg. Figure 30 shows a photograph -50- f & ORNL—-LR—DWG 22434 REYNOLDS NO. SEPARATION POINT (DISTANCE FROM INLET) {in.) 3000 2 g 5000 2 Y 7000 24 10,000 1%/ 15,000 146 , 20,000 1% 30,000 1% —=— SEPARATION STARTS AT ABOUT THIS POINT AND ENDS AT THE EQUATOR OR JUST BELOW IT | \\ LOW NEGATIVE FLOW NEAR THE \ SHELL WALL ¢ 1 | | } —w—LOW FLOW NEAR THE SHELL WALL BUT NOT NOT NEGATIVE Fig. 26, Qualitative Velocity Distributions at Nee mid = 3,000 and General Flow Features at Midplane Reynolds Numbers up to 30,000 for 1.443:1 Constant-Gap-Width Core Model. -5]- i ORNL—LR-DWG 22432 THICK BOUNDARY LAYER AT BOTH THE ISLAND AND SHELL WALL THICK BOUNDARY LAYER AT THE ISLAND /SEPAF\’ATION STARTS AT ABOUT THIS POINT \\ \ \ THICK BOUNDARY LAYER NEAR THE ISLAND \ | _— SMALL EDDIES NEAR THE SHELL WALL LOW FLOW NEXT TO ISLAND Fig. 27. Qualitative Velocity Distributions at Nee mid = 3,000 and General Flow Features at Midplane Reynolds Numbers up to 20,000 for 2,133:1 Constant-Gap-Width Core Model. -52- ORNL—LR—DWG 22429 Z’w/ INTERMITTENT REVERSE FLOW | 111 Fig. 28, Qualitative Estimate of Axial Component of Velocity Distribution at Axial Nee mid = 3,000 for Rotational Flow Through ART '"'18-Inch'' Core Model. I -53- $ii e ORNL—LR— DW! 22430 ANGLE OF FLOW REFERRED TO HORIZONTAL PLANE LARGE TURBULENT EDDIES NEAR ISLAND WALL ANGLE ~ O deg TO —20 deg NEAR SHELL WALL ANGLE ~ +50 deg DASHED LINE INDICATES SEPARATION REGION VERY HIGH VELOCITY NEAR THE SHELL WALL NEAR ISLAND WALL ANGLE ~ —40 deg TO ~-90 deg AVE. —-28 deg NEAR SHELL WALL ANGLE ~ +60 deg NEAR ISLAND WALL ANGLE ~ O deg TO —30 deg NEAR SHELL WALL ANGLE ~ +54 deg MANY EDDIES IN THIS REGION NEAR ISLAND WALL ANGLE ~ O deg TO +28 deg NEAR SHELL WALL ANGLE ~ +26 deg +29 deg Fig. 22. Qualitative Axial Velocity Component Distribution for Axial Neo mid™ 3,000 for Rotational Flow Through ART ''21-Inch'* Core Model. Re, mi ~54= Fig. 30. '"'21-Inch'' ART Core Model with Turning=Vane Section at the Inlet, G -55- of both the vanes and the core model. Flow Through 10/4U4-Scale ART "21-inch" Core - Twin Tangential Entrance Header - One-Pump Operation The gross flow features of a 10/blh-scale "21-inch" ART core with twin tangential entrance ducts (scaled from the ART medium rotationsl velocity entrance) were obtained by observing the motion of phosphorescent particles in the water while simulating single-pump operation. The flow features near the inner and ou;er walls are indicated in Figure 31. Arrows indicate the main direction of the flow near the walls. Flow Tprough Cores with Turbulence Promotion (a) Axial flow through the "18-inch" ART core model with twelve l6-mesh screens spaced one inch apart in the inlet pipe was Studied16 using the phosphorescent particle flow visuslization technique. Figure 32 shows the arrangement of the screens and the core used in this study. Qual- itatively sketched velocity profiles are shown in Figure 33 along a vertical section of the flow channel. (b) The "18-inch" quarter-scale core model was equipped with eight aerofoil- type turbulence promoter vanes attached to the outer wall at the inlet.17 Figure 34 shows the vanes attached to the outer shell, and the three vertical planes where visual observations were made. The island has been removed to provide a clear view of the vanes. Qualitative velocity profiles in the three planes mentioned are shown in Figure 35 at N = 5,000, The gross flow features were also observed at Re, mid Ve, mig = 20,000, -56- UNCLASSIFIED ORNL-LR-DWG 22428 REGION OF VIOLENT EDDYING EXISTS NEAR INNER WALL, BE- COMING LESS VIOLENT AS THE DISTANCE FROM THE HEADER INCREASES SOLID ARROWS INDICATE GEN- ERAL FLOW DIRECTIONS NEAR INNER WALL, DASHED ARROWS INDICATE GENERAL FLOW DIRECTIONS NEAR QUTER WALL. FIRST VIEW OF CORE MODEL SECOND VIEW OF CORE MODEL (90° FROM FIRST VIEW) /FLOW IN —=—FLOW IN Q” TH¥RD VIEW OF CORE MODEL FOURTH VIEW OF CORE MODEL (480° FROM FIRST VIEW) (270° FROM FIRST VIEW) Fig. 31. Flow Features Near the Inner and Outer Wall of the "24-in." ART Core Model while Simulating Single-Pump Operation -57= ORNL—LR—DWG 22425 {2—16 MESH SCREENS PLEXIGLAS TEST SECTION Fig. 32. Sketch of Core Model Assembly Showing Location and Spacing of Screens in Inlet Pipe, ~58— ORNL—LR—DWG 22426 A Wy NEGATIVE PROFILES EASY TO OBSERVE. A Fig. 33. Qualitative Estimate of Velocity Distribution for Axial Flow Through ''18-Inch"’ ART Core Model with Screens at the Inlet Nre mid = 3,000, -59- e UNCLASSIFIEL ‘ - | PHOTO 23991 Observation Planes Between Convergent Pair Below Vane Between Divergent Pair ‘ i it ATy . Fig. 34. Plastic Outer Core Shell with Turbulator Vane Set No, 1. -60- ORNL-LR-DWG 22427 VANE POSITION POSITION ‘ BETWEEN DIVERGENT BETWEEN CONVERGENT VANES BELOW VANE VANES \\ A Y / o W \ \ \ i | ] N W LARGE, FLUCTUATING HORIZONTAL VELOCITY COMPONENTS \ OBSERVED IN SEPARATION REGION AROUND EQUATOR Il Y DOTTED LINES REPRESENT LESS FREQUENTLY SEEN PROFILES » -—_'____.——-7 | s’ ~ Y /'? ! / m ,1’? \\ / / j Ny Y W ] /) W Fig. 35. Qualitative Estimates of Velocity Distribution and General Flow Features Through the ''18=Inch'' ART Core Model with Turbulator Vane Set No. 1 at Midplane Reynolds Numbers from 5,000 to 20,000, —_— -61- (c) (d) (e) (f) Eight aerofoil turbulence promoters of sharper curvature and wider span were also tried.18 Observations of velocity profiles were made slong the vertical planes shown in Figure 36. This figure is a photograph of the core shell with the vanes attached. Figures 37 and 38 show the veloc- ity profiles obtained qualitatively at N o = 5,000 and the gross flow Re, mid features observed at NRe, mid = 20,000, Qualitative axial velocity profiles were obtained using the phosphores- cent particle technique for the "18-inch" ART core model with the Pratt and Whitney Aircraft vortex generator in the entrance.19 Figure 1% shows the generators mounted on the core, and Figure 39 shows a photograph of the generator. The vanes for this test were set at a 45-deg angle with the core axis. Figure 40 shows the qualitati#e velocity profiles obtain- ed. Average profiles are shown since variability existed in the flow. General flow features were also observed up to N = 20,000, Re, mid The Pratt and Whitney Aircraft vortex generators used in case (a) above were also tried with the "21-inch" ART core model.™” The axial velocity profiles are shown in Figure 4l. The dashed lines indicate other instan- taneous profiles observed at different times. The general flow features were observed up to NRe, mid = 20,000, Experiments were conducted on the "2l-inch" ART core model with Pratt and Whitney vortex generators of 50-, 55-, 60-, 65-, and 70-deg vane sngles relative to the direction of the core axis. Unsteady flow (large-scale fluctuating eddies) was observed using the phosphorescent particle flow visualization method for all cases. The results were very similar to b= UNCLASSIRJED PHOTO 23992 Observation Planes Between Convergent Pair Below Vane Between Divergent Pair Fig. 36, Plastic Core Shell with Turbulator Vane Set No. 2. FSTI Sin. -63- ORNL-LR-DWG 22423 VANE POSITION VERTICAL VERTICAL VERTICAL PLANE PLANE PLANE BETWEEN BELOW BETWEEN DIVERGING VANE CONVERGING VANES VANES 1 i | e N, e L | W W \ \ w | \ v\ | | ’ \ C 1 | AN | I | M [ oy ‘\ 1 - " \ -~ O\ Y s IR\ 7 | \ \ v \‘ \. ‘\ : : ~ A | I {7 N \ l - !\ | ! - ! _/m l‘ T r T W , L I~ | ! 7 ! | L W W ! : . A MIDPLANE Nge = 3000 LARGE-SCALE FLUCTUATIONS ARE VISIBLE IN THE CHANGING PROFILES AT CERTAIN STATIONS A N ———— MARKS EXTENT OF SEPARATION REGION IN VERTICAL PLANE BETWEEN DIVERGING VANES — —— MARKS EXTENT OF SEPARATION REGION IN VERTICAL PLANE BELOW VANE ——-~—MARKS EXTENT OF SEPARATION REGION IN VERTICAL PLANE BETWEEN CONVERGING VANES Fig. 37. Qualitative Estimate of Velocity Profiles Observed in Axial Flow Through the ''18-Inch'' ART Core Model with Turbulator Vane Set No. 2 at NRe mid = 3,000, b4 ORNL-LR-DWG 22424 BETWEEN DIVERGING BETWEEN CONVERGING VANES BELOW VANE VANES SEPARATION REGION — ) — — e — — —— s e Fig. 38. Qualitative Estimate of Separation Region in Axial Flow Through the ''18-Inch"’ ART Core Model with Turbulator Vane Set No. 2 at NRe d = 20,000, r N -l AN =, " wr agq.. -66- ORNL-LR-DWG 22424 THICK BOUNDARY LAYER BUILDS UP NEAR THE ISLAND AT APPROXIMATELY THIS LOCATION AND ENDS AT THE ‘ EQUATOR. THE PROFILES SHOWN ARE ONLY [ 1 AVERAGE SINCE THERE WAS SOME VARIABLE FLOW OBSERVED Fig. 40. Qualitative Estimate of Velocity Profiles Observed in Axial Flow Through the '"18-Inch'' ART Core Model with the Pratt and Whitney Aircraft Vortex Generator at the Inlet. L “'.‘-!» y % L i R ot -67- ORNL~LR-DWG 22422 DASHED PROFILES INDICATE THOSE THAT ARE LESS FREQUENTLY SEEN. PULSATING FLOW, VELOCITY NEXT TO THE ISLAND CHANGES FROM A VERY LOW TO A NEGATIVE FLOW. A'' SQUARE—WAVE" PRO- FILE WAS ALSO VISIBLE OCCASIONALLY. THICK BOUNDARY LAYER NEXT TO THE ISLAND Fig. 41. Qualitative Estimate of Velocity Distribution in Axial Flow Through the 1121 ~Inch' ART Core Mode! with the Pratt and Whitney Aircraft Vortex Generatorat the Inlet, -— -68- those obtained in case (e). In addition, periphersl asymmetries in the flow rate in the core were observed with the vortex generators, even though the distribution of flow to the vortex generators was even, (g) The inlet half of the ART "21-inch" core model was packed with combi- nations of screens of varying solidity ratios.eo A photograph of the core model with a screen packing combination is shown in Figure 43. The following table lists the screen combinations investigated: TABLE 4 Screen Packing Combinations for 10/44-Scale "21-inch" Core Distance from Combinations Position Core Inlet (in.) 1 2 3 4 5 1 0 a a 2 0.373 3 0.746 a & a a a 4 1.118 > 1.491 a a c a a 6 1.864 T 2.238 c a a 8 2.610 a c 9 2,982 a b »* a - 0.385 solidity ratio screen, 20 x 20 mesh, 0.0108-in.-dia wire b - 0.510 solidity ratio screen, 20 x 20 mesh, 0.0150-in.-dia wire c - 0.622 solidity ratio screen, 30 x 30 mesh, 0,0128-in.-dia wire D ¢ UNCL ASSIFIED PHOTO 27516 Fig. 42. 10/44-Scale Model of ART Core with Screen Combination Number 5. _— -70- (h) Figures 43 through 47 show the flow features and velocity profiles obtalned with these screen arrangements using the phosphorescent particle technique. The total relative pressure loss for the five combinations is shown in Figure 48 as a function of core midplane Reynolds modulus. It was found that the combination given by number 5 of Table 4 prevented boundary layer separation and created a fairly uniform velocity field. This is shown in Figure A47. The inlet area expansion of a 10/44-scale model of a "21-inch" straight <0 A sketch of annular core was also packed with combinations of screens. the flow channel with the screen positions indicated is shown in Figure 49 and a photograph of the core packing in Figure 50. Two combinations were tried: 0.385 solidity ratio screens in positions 1, 2, 3, 4, and 6; and 0.385 solidity ratio screens in all six positions. Figures 51 and 52 show the flow features of these combinations. Plotted in Figure 53 is the rel- ative pressure loss for the second packing combination as a function of core midplane Reynolds modulus. The second pecking combination prevented boundary layer separation in the abruptly diverging entrance region and gave a fairly uniform flow throughout the core. The boundary-layer sep- aration region shown in Figure 52 near the exit of the core results from the too abrupt turning angle in this region. Velocity Fluctuations in the Full-Scale "21-inch" ART Core Model Motion pictures21 of dye filaments near the walls in a full-scale model of the "2l-inch" ART core were analyzed frame by frame to determine the angle of, the dye filament at the point of injection. The results are plotted in -71- Ornl-Lr-Dwg-17718 BOUNDARY LAYER SEPARATION 0.385 0.385 0.622 | "lll BOUNDARY LAYER SEPARATION \ IN SMALL AREAS ABOUT | 'IIIV PERIPHERY SCREEN POSITIONS INDICATED BY SHORT ARROWS. NUMBER NEXT TO ARROW IS SOLIDITY RATIO OF SCREEN. VELOCITY PROFILES OBSERVED AT ReyvippLane = 7,650 PARTICLE MOTION OBSERVED UP TO RemippLane = 80,000, i FLOW EXIT Fig. 43. Qualitative Flow Distribution in ART Core with Screen Combination Number 1, =72~ Ornl-Lr-Dwg-17719 BOUNDARY LAYER SEPARATION 0.385 VELOCITY PROFILES OBSERVED AT RemippLane = 5000 SCREEN POSITIONS (NDICATED BY SHORT ARROWS. 0.385 NUMBER NEXT TO ARROW IS : SOLIDITY RATIO OF SCREENS. 0.385 BOUNDARY LAYER SEPARATION AT SOME POINTS AROUND PERIPHERY FLOW EXIT Fig. 44. Qualitative Flow Distribution in ART Core with Screen Combination Number 2. -73= Ornl-Lr-Dwi—l7720 BOUNDARY LAYER SEPARATION AT SOME POINTS AROUND PERIPHERY PARTICLE MOTION OBSERVED UP TO ReMIDPLANE = 80‘000 SCREEN POSITIONS INDICATED BY SHORT ARROWS. NUMBER NEXT TO ARROW IS SOLIDITY RATIO OF SCREEN Fig. 45. Qualitative Flow Distribution in ART Core with Screen Combination Number 3. ~74- Ornl-Lr-Dwg-17721 0.385 , SCREEN POSITIONS INDICATED “—0.385 BY SHORT ARROWS. NUMBER NEXT TO ARROW 1S SOLIDITY RATIO OF SCREEN. VELOCITY PROFILES OBSERVED 0.385 AT Rey ppLang = 6000. PARTICLE MOTION OBSERVED UP TO REmippLane = 80,000 -=—— (). 385 0.385 ~~————THICK BOUNDARY LAYER THICK BOUNDARY LAYER Fig. 46. Qualitative Flow Distribution in ART Core with Screen Combination Number 4. -75- Ornl-Lr-Dwg-17722 0.385 0.510 SCREEN POSITIONS INDICATED BY SHORT ARROWS. ' NUMBER NEXT TO ARROW IS SOLIDITY RATIO OF SCREEN. ) VELOCITY PROFILES OBSERVED AT 3OOO S N o @ ORNL-LR-DWG 22419 S':TATIOINS —C[)UTER’ WALLI l | [ | l 1 | ] l | | | 8 7 6 5 4 3 2 STATIONS—INNER WALL 8 ! 6 5 q 3 2 O OUTER WALL ® INNER WALL o e 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 DISTANCE FROM CORE INLET (in.) Fig. 55. Dye Filament Fluctuation Frequencies as Obtained by "Zero" Counting VELOCITY (ft/sec) VELOCITY (ft/sec) VELOCITY (ft/sec) 10 0 18 16 -85- ORNLQG 22448 TWO POINTS AT EACH POSITION REPRESENT TwO SEPARATE VELOCITY MEASUREMENT POINTS 90 deg. APART ABOUT CORE PERIPHERY HIGH VELOCITY ROTATIONAL FLOW 14 12 0 B8 ©6 4 2 0 2 4 6 DISTANCE FROM MIDPLANE (in 8 ) 10 12 L.OW VELOCITY ROTATIONAL FLOW-—GS-2 TURNING VANES . Vi 14 6 18 Va 18 16 14 12 10 8 6 4 2 O 2 4 6 8 10 12 14 16 18 DISTANCE FROM EQUATOR (in.) \\ MEDIUM VELOCITY ROTATIONAL ITLOW o \\ Ve = \‘\ v /V/ t \ / N o B rd ) o\ { 4 e T v iB 16 14 412 0 8 6 4 2 O 2 4 6 8 10 12 14 16 18 Moduli. DISTANCE FROM MIDPLANE (in) Fig. 56. Mean Axial and Rotational Components of Velocity in the "21- Inch” Full Size ART Core Modei for Water Flow at ART Design Reynolds DIFFERENCE, Ib/in? PRESSURE ORNL-LR-DWG. 12750 60 " \\ / / 30 \\ // 1 / © O~ \\-_E / I 20 CALCULATED PRESSURE DIFFERENCE F ' o o Y URGERSON'S DATA o o o 0 [ 0 4 8 12 16 20 24 28 32 36 DISTANCE FROM INLET - INCHES Fig. 57. Radial Pressure Difference Between Inner and Outer Walls for the High-Velocity Rotational Flow of Fluoride Mixture No. 30 at Design Reynolds Moduli in the ''21-Inch'' ART Core. ORNL-LR-DWG. 12752A 60 PUMP — ENTRANCE NO.2 (OUTER WALL) PRESSURE |— | o N _ —_—— 0 e s — — — — —— | o -_-_.___..--—-' "I—' _----- \O.\C ~—— - — INNER WALL (CALCULATED) ~ m --...______ N 40 c = N . Ll B —_— o s L ENTRANCE NO.1 (OUTER WALL) ~ 0 O x 20 — o 0— o 0= I — N - — —0— 0 \NNER WALL ENTRANCE NO.3 (AXIAL_FLOW \ WITH PERFO ED PLATE - INLET ' l ~—o0—" 0 0 4 8 12 16 20 24 28 32 36 DISTANCE FROM CORE INLET (inches) CORE INLET Fig. 58. Static Pressure Distributions in the ''21-Inch'' ART Core for Flow of Fluoride Composition No. 30 at Design Reynolds Moduli. F5 , Sy @ £ T FERA o e . - -88- friction loss, entrance region turning and eddy losses, and the radial pres- sure differences for the medium velocity rotational flow case. Reduced data for an axial flow case achieved by turning vanes followed by a perforated plate are also included. In all cases, the pump output pressure was taken s as 51.5 1b/in.2 (37 feet of fluoride composition No. 30). . 3 e L ovw DISCUSSION Literature Survey The dynemics of flulid flow through the ART and other proposed reflector- moderated reactor cores is complicated by the shape of the cores and their small length-to-hydraulic diameter ratio. A horizontal section through any of these core flow passages presented a ring-shaped, or annular, aspect. The area of this ammulus was described as some function of the axial distance from the inlet. Usually this area first increased and then decreased as the dis- tance from the inlet increased. A further complication wes added when the flow was introduced such that it proceeded in a spiral fashion through the core. The length-to-dismeter ratio of the cores was between 6 and 10. For spiral flows, the length of flow path to diameter ratio varied between 10 and 50, depending on the ratio of the spiral to axial velocity components. Although the whole flow problem is enormously complicated by these fac- tors, the problem can be clarified somewhat by considering a series of simpler problems. Disregarding the short L/b of the core and considering only axial flow, the hydrodynsmic field can be described as primarily turbulent flow in a divergent-convergent asnnular duct. Flow in Convergent and Divergent Channels. Consider Bernoulli's energy equa- tion for frictionless flow, v.g p v, 2 P,' 1 1 2 2 h, + 5@: + == h, + 26, - (23) Differentiating with respect to the direction of flow, x, there results, dh v dv _ _14dp' xteg &x Ty & (24) Since the body forces are balanced out by a corresponding pressure force, equation (24) can be written v d&v _ _14dp where %E is now the pressure change with respect to flow direction due to momentum changes (friction is not considered). Using the continuity equation for incompressible flow vA_ = constant = Q (26) av =d(9)=-—9—dA (27) A 2 "¢ c A c Substituting equations (26) and (27) into equation (25), Q c 192 ( — == 28) A3g dx 7y dx C c Thus, if the cross-sectional aree increases in the direction of flow, there is a corresponding pressure rise in the direction of flow. The assumption of frictionless flow leads to the conclusion that the veloc- ity distribution in a plane perpendicular to the flow is uniform at any plane; and, in diffusers, the diminution of velocity with distance from the inlet of the diffuser is uniform. These assumptions give a finite discontinuity at the walls. The walls do not move; and, at the wall, the velocity is that of the main stream., ‘}_:”'Q!-.: ’ = § However, fluids are not frictionless; and the equetions presented do not give the whole picture of flow through divergent channels. An interchange of momentum exists in all flulds that eliminates the dis- continuity at the walls so that the fluid velocity at the wall is zero, and the curve of velocity versus distance from the wall rises in a continuous fashion to the main stream velocity. This interchange of momentum gives rise to a "shear" force between the various layers of fluid which is expressed in the differential equation where %E is the pressure difference across the differentiel fluid element, between x and X + Ax as Ax —> 0; %% 1s the difference in shear stress across the element between y and y + Ay as &y —> 0, and éi %% is the change of mo- c mentum of the fluld in the element between X and x + AX as Ax —> 0, The shear force T, in laminar flow, is proportionel to the velocity gradient perpendicular to the flow direction; i.e.; T::“'d-_-&: . In turbulent flow, however, the shear stress is not proportional to the velocity gradient; i.e., T = 1 viu! gC Prandtl assumed that v' and u' are correlated to be of the same order of dv magnitude, thus both are proportional to 1 . E§ 3 and obtained 2 2 /dv 2 dv, dv dv T=rpl (dy) =l T S -— -2- v ooa where € has the dimensions of a viscosity coefficient, but is not constant. It depends on the turbulence intensity (v' and u'), indirectly. It can be seen, in general, that if the adverse pressure gradient (pres- sure rise in flow direction caused by the area increase) is greater than the shear gradient, the fluid element loses momentum. Since the fluid near the wall i1s retarded by friction, it has much less momentum than does the rest of the streem. The adverse pressure gradient (being uniform across the channel), if sufficiently large, can cause the velocity near the wall to taken on nega- tive values. Thus the forward flowing stream, in effect, separates from the surface, It is also noted that if the velocity distribution is made "flat" such that large velocity gradients occur near the walls, the shear stress gradients at the walls will be larger; and the fluid can force its wey against a larger adverse pressure gradient without separation. In flows in places other than in entrance regions, where the velocity is uniform to begin with, the "bound- ary layers"; i.e., layers of fluid moving;more slowly than the main stream due to the action of the stationary walls through the viscosity (laminar or tur- bulent) of the fluid, have grown in thickness such that the flow is substan- tially nonuniform. Here, the velocity distribution can be made more uniform by mechanically increasing the turbulence level somewhat uniformly across the channel; thus increasing €, which tends to equalize velocities in adjacent layers of fluid in the channel. Experiments have been carried out in smooth, two-dimensional channels of rectangular cross section with varying angles of divergence between the two opposite walls which form the long sides of the rectangular cross section.8 Horae o - -93- The flow into the channels was essentially uniform. The velocity distributions obtained by Nikuradse in this study for established flow through channels of various divergence angles (negative angles refer to channels converging in the direction of flow) are shown in Figure 59. No backflow regions exist at either wall in Figure 59-a; separaticn of the forward flow is just beginning to take place at 10 degress in Figure 59-b; at 12 degrees in Figure 59-¢, separation is more evident; and in Figure 59-d4 at 16 degrees, there is a tendency to switeh from wall to wall, each configuration being maintained for some time. At still larger angles, the separation takes place at bocth walls. In Fig- ure 59-a, it is also seen that as the divergence angle increases from O to 4 degrees the velocity tends to peak more sharply in the center of the channel; conversely, as the wall angle decreases from O degrees (converging channeis), the flow profile is flattened out. The maximum value of the area gradient obtained before separation of the forward flow took place was found to correspond to an included esngle between the divergent walls of 8 to 9 degrees. Similar experiments done with "dif- fusers" of circular cross section alsoc have roughly established the maximum included conical angle to be 8 to 10 degrees. In gomparison, the included angle of a conical diffuser roughly equivalent to the divergent half c¢f the "21-inch” ART core is ~30 degrees. From this, it is evident that separation ¢f the forward flow from the walls of the core will occur by virtue of the core geometry alone. That this separation of the forward flow from the walls is due to the adverse pressure gradient rather than tc any other property is demonstrated YU mnax VELOCITY {(cm/sec) (ag) 10 “8 Lpumrent™] 0.9 // -4 ///// -2 M P 1 o6 (05 !/ //a= 4 deg 0.5 ////, . 0.4 {/ 0.3 / a = HALF ANGLE OF CHANNEL 0.2 01 0 1.0 0.8 0.6 0.4 0.2 O v/t (c) 180 AT T 160 7 \\ 140 /' \ 120 \\\ 100 \\\ 80 60 40 \ 20 N\ a = 6 deg \ 0 ™ N -20 -40 -4.0 -2.49 ~-0.8 0.8 24 40 DISTANCE FROM CHANNEL CENTER (cm) -94- VELOCITY (cm /sec) VELOCITY (¢cm/sec) 160 140 120 100 80 60 40 20 O -3.2 280 240 200 160 120 @ O N O O | B @ -80 —-120 -160 - o] i UNCLASSIFIED ORNL-LR—-DWG 22417 // a = Sdeg / —1.6 0 1.6 DISTANCE FROM CHANNEL CENTER {(cm) {d) 3.2 / - \\ AN - // \\ a = 8 deg ' \\ o -5 -4 -3 -2 -1 0 1 2 3 4 DISTANCE FROM CHANNEL CENTER (cm) Fig. 59. Velocity Distributions in Convergent and Divergent Channels (Reference 8). 5 —-— -95- by an experiment conducted on flow through a pipe of rectangular cross section constructed such that the width of two opposite sides increased while the width of the other two sides decreased, the area remaining constant along the pipe. When fluid flows through such a pipe, there is no separation of flow from the wallsee. The "2l-inch" ART core also decreases in cross-sectional area from the midplane to the exit at the same rate as the area increases in the upper half such that the core is symmetricel about the "equator" or midplane. If separa- tion of the flow takes place in the upper half of the core, it becomes impos- sible to describe quantitatively the flow distribution in the lower half of the core without resorting to experimental measurements. Therefore, no con- clusions can be drawn from convergent channel datas as to the flow distribu- tion in the convergent half of the core. Flow in Curved Channels. The effects of spiral flow through ART and other reflector-moderated reactor cores can be seen by studying the simpler case of flow in a curved channel, When curved flow is obtained, body forces due to the change in flow direction play a role in the turbulent mixing., Thus, to balance the centrifugal force, a pressure distribution is set up across the flow which decreases the turbulent momentum exchange near the convex inner wall and augments the momentum exchange near the concave outer wall. Experiments have been conducted by Wattendorf23 on the turbulent flow of air in a two-dimensional curved channel. The velocity distributions for fully- developed flow in curved channels of outer radius to width ratio of 5:1 and 10:1 are compared to fully-developed turbulent straight channel flow in Fig- ure 60. It is seen that the differences between straight flow and curved flow -94- UNCLASSIFIED ORNL—LR—DWG 22416 16 T T 1 STRAIGHT CHANNEL .4 I R R CURVED CHANNEL CURVED CHANNEL .2 ——(2t/r =1 A\ (2 t/r ="1/) \\ 1.0 ’/ -——_?\ . N\ 0.6 0.4 0.2 0 10 08 06 04 02 O 02 04 06 08 410 INNER CONVEX WALL v/t OUTER CONCAVE WALL Fig. 60. Comparison of Fully—Developed Velocity Distri— butions in Straight and Curved Channels. (Reference 23) 6 5 \ CURVED CHANNEL——"\ 4 (2//r= 1/g) M - NEHEA x 3 i _ 5 s~ STRAIGHT CHANNEL T~y \\ ®ley 5 // \I ' ‘/ \ / ] \\ 1 e T CURVED CHANNEL \ V (2 4/r=1/g) . N 1| 10 08 06 04 02 0 02 04 06 08 10 INNER CONVEX WALL y// OUTER CONCAVE WALL Fig. o4, Comparison of Eddy Diffusivities in Straight and Bl g Curved Channels. (Reference 23) % [ -_— -97- become more pronounced as the ratio of channel radius of curvature to channel width is decreased. The eddy diffusivity distribution for the 5:1 curved channel is compared with that of a straight channel in Figure 61. Two values for the eddy diffus- ivity can be obtained because of the difficulty in deciding whether e + Rl< e = (29) VA vhich arises from Prandtl's reasoning based on the exchange of angular momentum, or e = (30) which is the generalization of the equation for laminar flow where Toc [V _ ¥ P dr T 23 Shear stress measurements by Wattendorf is the valid equation. give diffusivities lying between those given by the above expressions, being much closer to equation (30) than to equation (29). It is seen from Figure 61 that the eddy diffusivitiés are less than the straight channel diffusivities near the inner convex wall and greater, near the outer wall. As the ratio of radius of curvature to channel width increases, the cen- trifugal forces decrease for a given flow rate. Thus for ratios above 5:1, the -eddy diffusivities will fall somewhere between the straight-~channel and the -_—— -98- ¥ 5:1 curved-channel values. Since the curvature parameter is ~2:1, on the average, for rotational flow in the "21-inch" ART core, the differences be- tween straight and curved flow will be even more pronounced than shown by Figure 61, Flow Through Screens. Investigations by Baines and Peterson2h of flow through screens show that the relative pressure loss through a screen is largely a function of the screen form (single and biplane lasttices, round bars, perforated plates, woven-wire, etc.) and solidity ratio (the ratio of the ob- structed area to the total area of the screen). Below a Reynolds modulus of 1,000 based on the wire diameter of the screen, the pressure loss is also a weak function of Reynolds modulus, increasing slightly as Reynolds modulus decreases. Figure 62 shows the relative pressure loss as a function of screen form and solidity ratio for NRe, o > 1,000. Low solidity ratio screens tend to flatten any arbitrary velocity dis- tribution, whilé screens above a solidity ratio of ~~0.5 may magnify or create velocity profile asymmetries since small differences in hole sizes can cause very large differences in the pressure drops through various parts of the screen. Figure 63 shows the effects of screens of different solidity ratios on an arbitrary velocity profile. The intensity of turbulence at distances greater than 10 wire diameters downstream from the screen is a function of wire size and distance from the screen. Therefore, a large diameter wire will create a larger velocity fluc- tuation (i.e., intensity of turbulence) at a given point downstream from the screen than a small wire. The fluctuation frequency will also be lower for igg Larger wire. 99— Ornl-Lr-Dwg-17724 UNCLASSIFIED 100 T 80 |- CURVE 1 C. FOR THIN-PLATE ORIFICES ll ] - (PERFORATED PLATES) ] ] 60 [ CURVE L 2gAp _ 1 ¥ :/ 40 | WHERE C¢ = | 7 CURVE T SQUARE-BAR LATTICES / / -~ CURVE I¥ ROUND-BAR SCREENS I / . /// / / / / 10 /771 7 8 // 1/ / 1/ 6 /7/ V/AVA 4 7 / 2gAp / // v 2 f P Vo 1 N/II Jr 2 / ! 4 / ,/ // /I 0.8 1T 0.6 / / / / / ! / / 0.2 1 / / / / 0.4 / 0 04 02 03 04 05 06 07 OB 059 Fig. 62. Relative Pressure Loss as a Function of Screen Form and Solidity Ratio. (Reference 24) SOLIDITY RATIO (S) 1.0 -100- UNCLASSIFIED ORNL—-LR—DWG 22453 l 7 A 1.0 ‘ [ 0.8 — ¢ UPSTREAM DOWNSTREAM @ 0.6 X — —DOWNSTREAM 0.4 |— (PREDICTED) 0.2 E s =075 O ] 0 0.2 0.4 06 0.8 u/ug (@) /8 y/8 (c) y/B (d) s = 0.234 0.2 0.4 0.6 0.8 u/uq 1.0 1.0 1.2 1.2 1.4 Effects of Different Solidity-Ratio Screens on a One-Directional Velocity Variation. 0.2 0.4 1.2 1.4. Effect of High Solidity—-Ratio Screen on o Uniform Velocity Distribution. Fig. 63. Effect of Screens of Different Solidity Ratios on Arbitrary Velocity Distributions (Reference 24) Ty -101- Screens in Diffusers. The equalization of velocity distribution and the tur- bulence promotion effects of screens have been shown to prevent separation or 25 to restore unseparated flow in a diffuser. Screens prevent separation by a combination of a number of factors: (1) increasing the velocity gradient at the wall, (2) increasing the turbulent mixing and, consequently, the shear stresses near the wall downstream from the screen, and (3) reducing the ad- verse pressure gradient along the wall. In studies by Schubauer and Spangen- 25 wide-angle diffusers were "filled" by a number of screens spaced along burg, their length. In general, the number of screens required to "fill" the dif- fuser increased as their solidity ratios decreased. Figure 64 demonstrates the effect of a single woven-wire screen in a conical 30-deg included angle diffuser. -102- iz UNcL AsSIFEED ORNL— LR— DWG 22452 A,B,C, AND D REFER TO |PROPORTION OF TRAVERSE STATIONS AL SEPARATED REGION 0 0.8 9 0.6 q, 0.4 0.2 16 12 8 4 0 4 8 {2 16 -0.4 0 04 08 12 1.6 2.0 r,in. X /0q POSITIONS OF A,B,C, AND D ARE SHOWN AT RIGHT A,B,C,D, AND E REFER TO TRAVERSE STATIONS R 50-MESH SCREEN; 9.4 K=1.40 .2 r ITh o 7 Dy 04 18.0 0 -04 16 12 8 4 0 4 8 12 46 r,in. POSITIONS OF A,B,C,D, AND E ARE SHOWN AT RIGHT A,B,C AND D REFER TO TRAVERSE STATIONS 75-MESH BOLTING CLOTH; K=2.95 a4 q, 2.0 2.4 16 12 8 4 0 4 8 12 46 X /D r,in. 0 POSITIONS OF A,B,C, AND D ARE SHOWN AT RIGHT 75-MESH BOLTING PROPORTION OF TOT 0.8 |CLOTH; k=3.04 FLOW |a,B,c,D, AND E REFER T0 VERSE STA q I 0.8 9 Do 0.4 o8 0.2 c 0 6 12 8 4 0 4 8 12 46 -04 0 04 08 1.2 16 20 - in X/Dq POSITIONS OF A,B,C,D, AND E ARE SHOWN AT RIGHT Do =DIAMETER OF EXIT PIPE 2gAp R=WALL RADIUS . K=RELATIVE PRESSURE LOSS OF SCREEN v 5 # —DvNAMIC PRESSURE — % r=RADIUS ° = REFERENCE DYNAMIC PRESSUR x=DISTANCE FROM DIFFUSER ar URE ENTRANCE Fig. 64. Effects of Screen Placed Across a Conical 30 deg. Included Angle Dif- fuser. (Reference 25) e -103- Discussion of Results Axial Flow Through ART Core Models. It was seen from the results of the ex- periments on the quarter-scale "18-inch” ART core model and the 10/44-scale "21-inch" ART core model with a radially and peripherally uniform axial in- let flow that axial flow was always accompanied by boundary-layer separation and reverse flow in the separation region. The separation region began at a plane 0.7 in. below the inlet plane of the "18-inch" model and covered the outer wall to a plane approximately three-fourths of the core length from the inlet. The separation region in the "21-inch" model was approximately the same size, beginning about 0.5 in., below the inlet plane. It should be emphasized that the reverse flow in the separation region is violently unsteady, being composed of large vortex-like eddies and flow stagnations which were intermittently created and destroyed. These instabil- itiés, in turn, created unsteadiness in the main part of the flow adjacent to the separation region. Because of the short length-to-diameter ratio of the cores (~10), this unsteadiness was carried practically undiminished out the exit of the core by the flow. The boundary-layer separation observed results from the large adverse pressure gradient (pressure rise in the direction of flow) caused by the velocity decrease as the fluid moves into the core. In turn, the velocity decrease is caused by the increase in the flow area through the inlet half of the core. The comparatively lower fluid shear stress near the outer wall than the inner wall (common to flow through annular ducts) and the curvature of the channel away from the axis at the entrance probably caused the preferential lé« L 'IIIIII' L -104- g h separation from the outer wall. Axial Flow in Two Constant-Gap Core Models. The investigation of axisl flow through two constant-gap core modelslof midplane-to-inlet area ratios of 1.443:1 and 2.133:1 revealed that boundary-layer separation occurred at the outer wall in both cases. For the 1.443:1 area ratio core, the separation region was very thin and the point of separation moved toward the inlet as the flow Reynolds modulus increased. At a core midplane Reynolds modulus of 20,000, the separation slternately occurred at the inner and outer walls; while at 30,000, separation occurred only at the outer wall. With the 2.133:1 area ratio core, the separation layer was thicker on the outer wall and began at a point closer to the inlet than in the previous core. Very little depend- ence of the separation point on the flow rate was noted. These results verified the previous prediction that boundary-layer sep- aration would occur in axial flow through cores of midplane-to-inlet area ratios greater than 1.41:1. If the 1.443:1 core had been constructed so that its rate of area increase was similar to the 1.41:1 area ratio core, instead of being zero at the inlet and midplane and reaching a meximum value half-way between, separation might have been avoided. Rotational Flow Through ART Cores. It was found from the quarter-scale "18-inch” core model and 10/44-scale "21-inch" core model tests and from data obtained with the full-scale ART core model tests6 that rotational flow was always accompanied by boundary-layer separation and violently unstable re- verse flow in the separation region. The separation region in all cases be- gan at the inlet on the inner wall and extended approximately three-fourths of Fhe_core length from the inlet. CE A fi;:&h" r"“v - ‘i — _—_— ~105- o The separation here was also caused by the same adverse pressure gradient as before. The preferential separation from the inner wall was probably caused by the comparatively lower fluid shear stress near the inner wall than near the outer wall (a phenomenon of flow in curved channels). One-Pump Operation - ART Inlet Header. The same gross flow features described in the previous section were observed in one-pump simulated operation of the ART inlet header-"2l-inch" 10/l4-scale core model combination. In addition, a region of violent eddying was observed at the inner wall directly under the jet of fluid issuing from the tangential entrance. Peripheral flow asymmetries in the entrance half of the core were noted. The observations of the position of the eddying agreed roughly with the position of a large temperature asymme- try measured in the core volume-heat-source experiment run under the same con- ditions.27 In this study, the positions of the tangential inlets are reversed from those in the ART. However, this only results in a reversal of the flow struc- ture without affecting any of the flow details. Flow Through Cores with Turbulence Promotion. Three methods of turbulence pro- motion were tried in the "18-inch" and "21-inch"” ART core models. This approach to the problem of improving the core flows was suggested by the elementary con- siderations presented previously. If the amount of turbulent momentum inter- change in the flow can be increased between the low velocity boundary layers and the high velocity fluid in the midstream, separation could be eliminsted. In terms of fluid shear stress, this would mean that the turbulent shear stress gradient near the wall would be increased and thus, the forces on a fluid ele- ment near the wall would tend to move the element forward with the flow. T _ It was found in the experiments that the scale of the promoted turbulence (a definition of the average size of the eddies) was also important. For ex- ample, twelve 1l6-mesh woven-wire screens placed in the entrance pipe of the "18-inch" core model (Figures 32 and 33) added sufficient turbulence to the flow to move the point of separation farther downstream from the core inlet than when no screens were present. A striking calming effect was also noted. In axial flow through the "18-inch" core model without screens, the flow in the separation region was very unsteady; while with the screens, the reverse flow was much more steady and streamline in character. When turning vanes were added to give a rotational component to the flow, no differences were noticed either in the velocity profiles or in the general character of the flow for the cases of screens located before the turning vanes and no screens. It should be pointed out that the turning vanes add turbulence of a larger scale to the flow. Contrasted with this was the flow obtained using vortex generators of the aerofoil type attéched to the outer well at the inlet (see Figures 34 through 38). Though the separation region on the outer wall was not eliminated in these cases, peripheral flow asymmetries and variable horizontal velocities of low frequency were noted in the separation region. Since these flow fea~ tures had not been observed in axial flow through the "18-inch" core model, they were probably the result of the large-scale vortices generated by the aerofoils, The flow obtained in the "18-inch" and "21-inch" core models with several vortex generators (see Figures 14 and 39 through 41) of various turning-vane angle settings was also variable in nature, the "2l1-inch" core exhibiting flow q‘flfl&fifi' 'Il!!! ay -107- §¥eandl variability to a much greater degree than the "18-inch" core model. Eddyi y ying and periods of flow stagnation were visible at intervais in both cores. Tt is seen from these experiments that the induced turbulence must be of a small scale (corresponding to a high-frequency spectrum of veloclity fluctua- tions) and of high intensity (large ratio of fluctuasting velocity component to the mean velocity). If the scale of the turbulence is large enough, the fre- quencies of the velocity fluctuations are low enough to be called varlable flow. Since turbulent flow is unstable flow, the question arises as to where one draws the line between turbulent flow and variable flow. In connection with high-power, high-temperature circulating-fuel reactors such as the ART, the answer is provided by the restriction that the turbulent eddies must be of a small enough scale to insure that the associated wall-temperature fluctua- tions are of sufficiently higfi frequency and small magnitude to be considered negligible. This is the safest assumption to make, since the effect of low- frequency thermal cycling on the wall material is not yet fully understood. For turbulent flow through a parallel-plates system.e8 where the wall- temperature fluctuations are negligible from the low frequency, large magni- tude thermal cycling standpoint, the ratio of the characteristic scale of turbulence (perpendicular to the flow) to the channel width is.of the order of 0.1, while the ratio of the scale of turbulence (perpendicular to the flow) to the wire diameter of a turbulence-producing screen‘ej1L is of the order of 1, on the average. Therefore, in the ART model core channel of average width ~ 0.8 in. with screens of wire diameter ~0.02 in. (for the l6-mesh screens), fhe scale of turbulence produced by the screens is about one-fourth of the scalehgf turbulence in the flow through a parallel-plates system of equal .o el & A . channel width. The ratio of the root-mean-square lateral velocity fluctua- tions to the mean velocity of the flow is also much larger behind a screen (0.2 - 0.5 maximum) than the corresponding values for turbulent flow in a channel (0.03 - 0.04% maximum). The magnitude of random turbulent velocity fluctuations established a short distance behind a screen decay at a rate predicted by Frenkiel's decay law: - X\ 5/ x_ = 1,12 (_&_) for — > 2 . < e M Since the scale of the turbulence is proportional to the screen wire diameter, it 1s seen that smsll-scale turbulence decays more quickly than large-scale turbulence. From these considerations, it 1s evident that if the scale of the pro- moted turbulence is smdll - as it must be to prevent large temperature ex- cursions - the turbulence dies out quickly behind the turbulence promoting device. This was indiceted clearly by the striking calming effect of the screens placed in the inlet pipe on the flow through the "18-inch" core model. Evidence of the effect of promoted turbulence of & large scale on the temperature.structure was seen in the ART core volume-heat-source experiment27 with a half-scale model of the guide vane and baffle plate arrangement shown in Figure 65.29 While the boeundary-layer separation regions at the core wemlls were eliminsted by the action of the baffle plate on the flow, wakes consist- ing of vortices shed from the edges of the baffle plate and the trailing edges of the vanes created & low-frequency variable flow in the core. -_— - Figures 66 through 683° illustrate the type of flow occurring in wakes and separation regions. Figure 66 is a photograph of flow in a separation region (in this.case, behind a stalled serofoil), Figure 67 shows the wake behind a cylinder (repreéentative of a screen wire), and Figure 68 pictures the flow around a flat plate held perpendicular to the stream. Flow Through Cores Packed with Screens. Due to the high rate of turbulence decay in the flow downstream from & screen and the unsteadiness induced in the flow by large-scale turbulence promoters, it became apparent that turbu- lence promotion at the entrance piane of the core would not be satisfactory from the standpoints of flow separdation elimination and flow stabilization. The placement of screens in the core seemed to be the logical answer to the decay of the promoted turbulence. The work of Schubauer and Spangenburg25 noted earlier showed that when screens of low-solidity ratio (< 0.5) are placed across a wide-angle diffuser, not only was turBulencepromotea by the screens; but the velocity distribution across the flow channel was made more uniform near the screen, If a sufficient number of screens were placed across the diffuser at different locations along the axis, boundary-layer separation was completely eliminated and the flow was made more steady. | The velocity redistribution effects which resulted from the screens placed across the diffuser cen be explained by first considering a plane jet?impinging against a solid well perpendicular to the jet axis. As the well is afiproached, the streamlines of the jet bend away from the axis and finally become parallel to the wall, the streamlines describing a family of hyperbolic curves. ART Fig. &5. GS—2 Guide Wanes and Baffle Plate Mounted on Inner Woll of Core ot the Inlet, |Reference 29) Fig. eT. Visuglization of the Flow in the Wake o F o o Cylinder, (Reterence 30) the 24-inch A \ n\ \ } \\ \ | ) I' o r 1l 8 | / j / Fig. 66. Visualization of Flow in o Separation Region Behind on Aerofoil. (Reference 30) r UNCL LSSIFIED PHOTO 004 Fig. 68. \isuolization of the Woke Behind o Flal Plate Held Perpendicular fo the Stream, (Reference 30) “0llL- —— 111 . Centrifugal pressure gradients accompanying the curving flow create a pres- sure distribution which increases toward the jet axis and toward the wall. According to Bernoulli's law of energy conservation, the fluid velocities decrease as the pressure increases. If the wall is porous, some of the fluid passes through the wall while the rest is turned as before. The extent to which the fluid is turned decreases with increasing porosity; or, in terms of the solidity ratio (the opposite of porosity), the stream passes through the wall with more and more of its original concentration as the solidity ratio decreases. Thus, a screen placed across a diffuser forces the high kinetic energy fluid away from the axis of the stream, where it tends to concentrate, toward the walls; and a flattening of the velocity profile occurs in the vicinity of the screen. The degree of flattening depends on the relative pressure loss of the screen, which is a function of the solidity ratio, and the form of the screen (square bars, round bars, single or biplane lattices, woven-wire screens, perforated plates, etc.). The more uniform velocity distribution plus the tur- bulence generated by the screen help to move the fluid near the wall behind the screen in the direction of flow, thus preventing separation of the flow. The "filling" effect of a screen in a diffuser evidently exists only in the vicinity of the screen, the extent of the vicinity depending upon the rate of expansion of the diffuser (large rate of expansion corresponds to a small extent of the vicinity) and the relative pressure loss (large pressure loss corresponding to large extent of influence), It was found that five screens arranged as given by combination No. 5 of Table_&lfrevented separation and gave a fairly uniform flow in the ART core. o _—— 12- S In the inlet area expansion of the straight annular core, six 0.385 solidity ratio screens gave a fairly uniform flow. It is seen that approximately the same number and type of screens were packed into the wide-angle divergence of the straight core as were spread over the upper half of the ART core to pre- vent separation. This bears out the considerations presented above. 1In addition, when & 0.510 solidity ratio screen replaced the farthest downstream 0.385 solidity ratio screen (combination No., 4 of Teble U4), the resulting velocity distribution following the screen was more uniform. (See Figure 47.) This is agein in accord with the above discussion. In the straight core, the boundary-layer separétion region seen in Figures 51 and 52 is a result of the abrupt curvature of the channel near the exit. The pressure must rise a;ong the outer wall and fall along the inner wall as the curve is approached so that a radial pressure gradient is obtained to balance the centrifugal force generated by the curving flow. The adverse pressure gradient, or pressure rise in the direction of flow, along the outer wall was apparently too large; and boundary-layer separation was the result,. If the separation can be eliminated by increasing the radius of curvature of the flow path, or by eliminating entirely the flow area contraction at the exit, this core model may possess a number of advantages over the present ART core with screens. These are as follows: (a) The screens, being in a lower neutron flux region, would absorb fewer neutrons, thus having less effect on the re- actor criticality. S 13- §orw (b) The screens are concentrated in the coolest part of the reactor, easing the problem of screen corrosion. (c) The volume of the core is larger (3.89 £43 compared to 3.23 ft3) reducing somewhat the average power density. (d) The fuel channel width is smaller (a maximum of 4 in. compared to 5-1/8 in.), causing a reduction in the channel wall-to-centerline power density ratio due to reduced self- shielding. The last two considerations would reduce the uncooled wall to mixed-mean tem- perature differences in the fuel. A problem may exist in adequately cooling the screen packing under zero fuel flow conditions and at times when the fuel is dumped from the reactor. Investigations are indicated to specifically ascertain the extent of this problem. A perforated-plate type of screen would have the same effect as the woven-wire screens on the flow through the cores, i1f its relative pressure drop and web thickness were the same as the relative pressure drop and wire diameter of a woven-wire screen. The solidity ratio of this "equivalent” perforated plate can be found by referring to Figure 62, find the pressure drop assoclated with a woven-wire screen of a particular solidity ratio on Curve IV; and then, using that pressure drop, obtain the perfeorated-plate solidity ratio from Curve I. For structural and fabrication reasons, the perforated-plate type of screen may be more desirable than the woven-wire screens used in the experiments. Velocity Fluctuation Studies. The fluctuation "frequencies" were obtained — A14- from the movies taken in the dye filament studies21 with the full-scale model of the ART core. The analysis was accomplished by counting the number of times the angular deflection curve crossed the average angle curve and divid- ing by the elapsed time (about 1.5 sec). It will be noticed that the fluc- tuations have no obserable periodicity; they are completely random. No con- clusions could be drawn from the magnitude of the fluctuations, since the distance of the dye filament from the wall was not exactly known; even though the angle measurements were made at the point of injection and the dye fila- ment was known to be close to the wall at that point. The components of the velocity fluctuations influencing the dye filament were also unknown. However, the frequencies were taken to be the frequency of changes of conditions at the walls, and the fundemental mode of the frequencies was in general agreement with the fundamental mode of the temperature fluctuations a7 in the volume-heat-source experiments after allowance for the difference in average velocities. It is felt that, given the same conditions, the fluctustion frequencies are proportional to the flow rate. This was qualitatively ascertained by com- paring movies of dye filament fluctuations for two different flow rates in the same core. Average Fluid Velocities in the Full-Scale "21-inch" ART Core. It is observed in Figure 56 that the data are in fair agreement with the predicted average velocities except for the high~velocity rotational flow case. Here, the rota- tional velocities fall below the prediction. Since the axial average velocity components were in fair agreement with predicted values and since the exit . 5 % 5 o e T L 13- B 4 rotational velocity profiles at two peripheral stations 90 deg apart prac- tically coincided, it was concluded that the high frictional drag had created a noticeable "vortex decay", or slowing down of the rotation, as the fluid passed through the core. Frictional effects were not considered in the pre- diction. This was further substantiated by the increasing difference between the predicted and experimental values as the core exit is approached. A backflow occurred near the exit on the inner wall in the high-velocity flow which also indicated the vortex decay, the backflow being induced by the adverse pressure gradient created by the deceying rotational component of velocity. It was observed that a large discrepancy existed between the predicted and experimental values for the average rotational flow component in the en- trance half of the core for the high- and medium-velocity swirl cases. This was noticed because the angular momentum of the flow in this region increased as the equator was approached, which is impossible without an external force acting upon the fluid. The conclusion was reached that the data are erroneous in the upper half of the core for these two cases. This conclusion is further supported by the fact that due to separation of the flow, a large amount of cross-flow exists in the upper half of the core. This would cause errors in the data which were obtained by impact "claw" probes. Since the probes could not be rotated to face the cross-flow, they would tend to read low; and since the impact tube and "yaw" tube arrangement was not calibrated in cross-flow, an erroneous vaelue for the flow angle could also result. \’ " “ ' . » AL %&% LRy ‘}' -— 116~ o Further support for this conclusion comes from the vaned-flow case in Figure 56. Here, separation and the accompanying cross-flow were probably eliminated, and the flow was primarily in the plane of rotation of the "claw probe. It is seen that no such discrepancy exists in the rotational component and that the trend of what little data there is follows the predicted rotation- al flow curve, Pressure Distributions in the "21-inch" ART Core. The vortex decay in the high-velocity rotational flow also explained the difference between the pre- dicted and actual pressure differential between the inner and outer walls for that case. The prediction was based on the predicted velocity distribution. In the case of the medium-velocity rotational flow, however, the prediction of average velocities agreed well with the data; so that the prediction of the pressure distribution on the inner wall using the outer wall pressure data and the calculated difference is probably sound. The pressure distributions shown in Figure 58 for the three cases (high- and medium-velocity rotational flow, and axial flow) showed that the total pres- sure loss was large for the first case (about 45 psi) and relatively small for the other two cases (about 7 psi). In the case of the axial-flow system, most of the pressure loss was caused by improperly designed turning vanes and a per- forated plate placed at the core entrance. Pressure Distributions in ART Heads. Since the pressure loss through the core is small, additional disturbances can be caused by pressure unbalances in the core entrance header., For example, the pressure loss was 2 1_b/in.2 at a 1200 gpm fuel flow rate for the screen-packed core shown in combination No. 5 3 ¥ - -117- &‘ e | of Table 4 and less than 2 psi without screens (see Figure 58). The calculations showed that the header shown in Figure 20(c) gave the smallest pressure unbalance, though even this unbalance was half the core pres- sure loss with screen packing. In an open core, these unbalances would cause a considerable amount of peripheral flow, adding to the unsteadiness already present, Since the velocity distribution in the exit half was evidently nearly uniform with periphery (see Figure 56), the cross-flow probably took place in the inlet half of the core. The actual existence of such cross-flow was masked by the low accuracy of the data in that region and the violently eddying sepa- ration region. With the large pressure differences found, the assumption of uniform fluid withdrawal from the header is no longer a good one; and the asym- metrical withdrawal of fluid will serve to lower the pressure rise somewhat. The calculations do point up two considerations, however. One is that the header pressure unbalances, for the configurations presented, are primarily due to the average velocity level in the header. Therefore, header fluid veloc- jties should be as low as practical. This is afforded by the header in Fig- ure 20{c). The core pressure loss should be as large as practical to keep the relative importance of the header pressure unbalances small. This was effected by the screen packing scheme, which also eliminates separation and destroys the thermal boundary layers in the core. - CapEie. -118- i CONCLUSIONS The following conclusions were reached as a result of this series of experiments: (a) Axial flow through the cores with sufficiently large cross-sectional area expansion rates considered here will always be accompanied by a separation of the forward flow from a point on the outer well near the inlet and violently unsteady reverse flow in the "separated” re- gion. This separation at the outer wall is due to the large adverse pressure gradient resulting from the large area expansion rate, the lower fluid shear stress at the outer wall than at the inner wall (common to flow through ennular ducts), and the curvature of the channel. The flow into the entrance of the core models was always uniform radially and circumferentially, having passed through a calm- ing length of 40 diameters of straight pipe into an annular "nozzle" which was mounted on the core entrance. (b) Rotational, or spiral, flow through the same cores is always accom- panied by separation of the forward flow from a point on the inner wall near the inlet and unsteady reverse flow in the separated re- gion with a velocity component in the direction of the rotaticn. The separation in this case is caused by the same adverse pressure gradient and the lower fluid shear stress at the inner wall relative to the outer wall (a phenomenon of flow in curved channels). Again, the flow at the core entrance was uniform radially and circumferen- ti?lly, having passed through 40 diameters of straight pipe intc an (c) (d) (e) (£) Ly T oA Lo annular "nozzle" which preceded the turning-vene section which pro- duced the rotational flow. In rotational flow, the turbulent interchange of momentum and the eddy conductivity are diminished at the inner wall due to the centrifugal force field set up by the fluid motion; and, therefore, increased tem- peratures due to volume heat generation can be expected at the inner wall over the temperatures obtained in straight flow at the same rate in an equivalent channel., If the spiral velocities are very high, frictional forces become of such magnitude that a decay of the spiral velocities becomes notice- able and a backflow due to spiral vortex decay will begin. This was noticed in the high spiral velocity case studied by Whitman, Stelzman, and Furgerson. Backflow occurred on the imner wall at the exit plane. Turbulence promotion within the cores has been shown tc overcome the effects of adverse pressure gradients, but the promoters must be such that they do not introduce unsteady flow; i.e., the scale of the pro- moted turbulence must be very small. Woven-wire screens and perforated- rlate screens of low solidity and small wire diameter {or web thickness) packed into the divergent part of the core flow channels are of such nature. Turbulence promoters such as vortex generators or large ob- structions placed in the core entrance cause unsteady flow although they seem tc eliminate the separation and the assocciated backflows. Straight annular cores and cores of sufficiently low expansion rates can be constructed to give a steady, unseparated flow fiith axial flow. Crlculations show that the maximum midplane-to-inlet area ratio that L K can be achieved with axial flow through a bare core is l.33:ljwithin an 18-in, lerngth of a 21-in.-0D core with the same midplane area as the "21-inch" ART core. | (g) Calculations have shown that a pressure unbalance (4 lb/in.e) due to momentum transfer exists in the header which is of the same order as the friction losses in the core. This pressure discrepancy is in the form of a rise in pressure as the fluid traverses the length of the header from the inlet duct. The unbalance, plus any unsteadiness trans- mitted to the flow by the fuel pumps, will also create peripheral flow asymmetries and unsteady core flow. The calculations also point up two considerations. One is that the header pressure unbalances are primarily due to the average fluid velocity level in the headers considered. Thus, header fluid veloc- ities should be kept as low as practicable. The core pressure ;oss should also be as large as practical to keep the relative importance of the header unbalances small. (n) Since screen packing in the core has been shown to eliminate the un- steady flow, which is an inherent characteristic of the core shape, it is proposed that this system be used in the core. An additional advan- tage accrues from the use of the screens in an increase in core pres- sure loss and the related velocity profile flattening. The peripheral asymmetry due to pressure unbalances in the header will then be much less than exists without the screens. It is felt that perforated-plate screens with the same relative pressure loss and mesh size as the wire o 121 .o screens tested would be more advantageous from the structural and fabrication standpoint than the wire screens. The heat-transfer characteristics of the screens are in the proc- ess of being investigated to determine whether any problems exist in this regard. (i)} A header system has been designed which may afford smsller pressure un- balances than the present header system and, being used in conjunction with the screen-packed core, may allow single-pump operation without large peripheral flow asymmetries. Further experimental work is in process to establish the validity of this conclusion and conc¢lusions (g) and (h). _—— -122- 4 APPENDIX A VELOCITY MEASUREMENT AND FLOW VISUALIZATTION TECHNIQUE Stroboscopic Particle Photography The messurement of velocities in the transparent "18-inch" ART core model was done by stroboscopic photography of the light scattered from a collimated plane of light parallel to the flow direction by tobacco-seed particles sus- pended in the water flowing through the core.ll The distance between two images of the same particle divided by the meas- ured time between the flashes of light which produced the images gave the aver- age velocity of the particle between the two positions., The particle velocity was very nearly the velocity of the water, since the density of the fiobacco- seed particles was close to that of water, and the particle size was smail. Since the curved surfaces of the "18-inch" ART core model cause an appre- ciable apparent distortion of the relative positions within the core model, a 1/16-in. thick Plexiglas sheet scribed with co-ordinate lines was first placed inside so that it lay within the plane of the light hbeam. This grid was supported by a second island which was split along its centerline and fitted with short pins to position a two-piece grid. The other half of the island was cut to fit the first with the grid halves in place, The plastic sheet was then edge-lit from two edges in a darkened room, the lines scattering light from the interior of the plastic and identifying their position; photographed by a 4 x 5-in. Grover monorail cemera equipped with a Kodak Ektar f4.7 lens facing at approximately right angles to the plane — 2 of the light beam; and removed. The resulting photograph is shown in Fig- ure 69. The camera was unmoved until after the particle photographs were taken so that the two negatives could be superimposed and the actuasl co-ordinates of the particles determined by reference to the similarly distorted grid system. The flash-light sources were General Electric FT-220 electronic flash- tubes operated by three 35-uf condensers in parallel and charged to about 2,360 v. A continuous light source was used so that the two images of each particle would be connected by a track, for identification on the film. This light was a General Electric H1000-B19, 1000-w mercury arc light. The light from all the sources was collimated by a pair of slits; the one farthest from the sources being common to all three sources, the sources being arranged to be in the same vertical plane. Figure 70 is a photograph of the test section with the split island and grid in place, the cemera in the foreground, and the collimator with the mer- cury vapor light operating. The side plate of the collimator has been removed to reveal the three slits in addition to the common one. One of the flash- tubes 1s in the cylindrical housing seen behind the upper collimator slit. The flashtubes were flashed and the time interval was measured by an electronic instrument consisting of an audio oscillator whose output frequency was manuelly controlled. A signal was sent from the oscillator to a conven- tional scaler whose output was one voltage pulse for 64 input pulses. The output of the scaler was further scaled down by two more stages in a special - ‘.h‘.ll-. R A - ool a4 J [ ! P 1 | | Bl i . &SI EE SEREA P REE - S i o= F . ------ [ L * 5 B ] e ] B SEE———— e g——— B S ah . . - __ — i 1 R L - [ L I e ..# i I ! | 1 - L e B = B I S [l - - I e B [ S I . = " i " . - P 141t _ ! b ! — | I I I e [ Ll 4 RN i G # o i L P e e - T L, HEE NI - i | I | | S R R R R o e F -125- W 7 LUNCLASSIFIED PHOTO 23825 okl [} ; ..; i.lll.-.l 1#“,,.':{..- _.-‘ : [ b ¢ R’ T R e . : R Fig. 70. Light Collimator, Test Section, and Camera Showing Relative Location. The continuous light source is operating. -— e "trigger finit." This unit then amplified and delivered the pulse to the first flashtube power supply which flashed its light. The second pulse, received 256 cycles from the sudio oscillator later, fired the second flashtube. The cycle was not repeated until the trigger unit was reset, When the flashtubes flashed, another signel was sent to a Hewlett-Packard Electronic Counter which was set so that the first pulse started it and the second pulse stopped it. The time between pulses was displayed on the counter front. Figure 71 shows a photograph of the electronic instrument. Kodek Tri-X Panchromatic and Royal Pan film were used in the camera for the recording of the light scattered by the tobacco seeds. Developing was done with Eastman DK-19a developer. However, it was found after the experi~ ments were finished that DK-19a increased grain size and fogginess in the film, and that DK-60a was recommended. A system of plane mirrors and a focusing lens and shutter were used to record the time displayed cn the counter face on the film in the camera. An over-all view of the apparatus is shown in Figure 72, The core model is all but hidden by the collimstor. To get;particle photographs, the core model was then reassembled with the original island, the connecting inlet piping put in place, and clear‘water with suspended tobacco seeds wes pumped through the core at a desired flow rate and temperature. The tracking light in the collimator and the automatic flashtube firing snd timing equipment were turned on. The procedure was then to open the camera shutter, trigger the automatic stroboscopic equipment, and close the shutter. The time displayed on the counter face was then photographed on UNCL ASSIFIED PHOTO 23827 Fig. 71. Electronic Flashtube Firing and Elapsed Time Measuring Equipment. -128- UNCLASSIFIED Y-12 PHOTO 23823 PLANE MIRROR I8 < um >~ # )) : F'L ANE MIRROR ' e = ¥ ‘h I TEST SECTION -l iR L . I e riul L B .'_ 2 . I ..f GE FT. 220 T e — \*‘" ‘ n POWER SUPPLIES ' ® Fig. 72. Over-All View of Test Apparatus G -129- the same film through the shuttered lens and mirror system. This procedure was repeated at least 100 times to obtain enough dsta to completely map the velocity field in the core. Typical data photographs are shown in Figures 73 and Ti. The velocity data obtained was shown in Figure 23. Phosphorescent Particle Flow Visuslization Flow visualization was accomplished by adding particles of Zn-CdS phosphor to the water in the system. Their motion, which closely followed the motion of the water, could then be observed by the light that they scattered from a colli- mated plane of light parallel to the flow from a continuous high-intensity light source, By creating a beam of light of short duration, either by mechanically "chopping" the collimated light from the continuous source (a high-pressure, 1000-w mercury-arc light), or by using the collimated light pulse from a high- intensity electronic flashtube (such as the GE FT-503), the phosphor particles in the flowing water in the path of the beam could be excited at a given in- stant, thus "tagging" them; and, a short time later, as the band of glowing particles deformed into a shape corresponding to the velocity profile of the fluid, they could be observed. A photograph of the mechanical chopper, which consisted of a solid ro- tating disk with a cut-out sector whose angle could be changed, mounted on the collimator is shown in Figure 75. From this method, a qualitative idea of how the velocity profile looked could be obtained. At the time of the core model experiments, no film was found to be sensitive enough to register the glowing particles so that, in a UNCLASSIFIED PHOTO 43045 Fig. 73. Typical Daoto Photograph. The small light spots at the lower left Fig. 74. Typical Data Photograph. The small light spots at the lower left ¥ are the numbers displayed on the Electronic Counter, are the numbers displayed on the Electronic Counter. -131- Sicssako PHOTO 27738 Fig. 75. Mechanical Light Chopper -— 192- manner similar to stroboscopic particle photography, the velocities could actually be measured. Recently, however, a feasibility study of photograph- ing the band of excited phosphors as they distorted into a shape correspond- ing to the velocity profile of water flowing in a tube was successful in ob- taining photographs.3l An experimental Eastman film with a speed of ASA L00-800 (S.0. 1177) wes used in a 4 x 5-in. camera with a £1.8 lens. Eastman DK-60=a developer was used. The method of making this method quantitative would be similar to the stroboscopic particle photography method; the time between the excitation of the particles and the taking of the photograph would be measured, and distances obtained by comparison with a photograph of a similarly distorted grid, which was previously placed in the plane of the motion of the excited particles. Velocities are obtained from the division of distance by elapsed time. _— 133- APPENDIX B FLOW SYSTEM COMPONENTS Reservoir A painted iron tank 40 in. deep and 24 in. in inside diameter served as the system reservoir. The top was open and afforded ready access for filling the system with water or adding particles for velocity measurement work. A motor-driven stirrer provided agitation to keep the particles sus- pended in the water. All connecting piping entered the reservoir under the level of the water maintained while the system was in operation, thus avoid- ing o’'r entrainment from that source. Centrifugal Pump An Allis-Chalmers double-intake centrifugael pump with a 9-in. diameter impeller driven by a 5-hp electric motor was used to pump water around the flow circuit. The pump delivered 85 gpm at a 80-ft head. At shut-off, a head of 84 ft was obtained. At a later date, a 10 3/441n. impeller was substituted and a T l/2—hp electric motor was employed to drive the pump. With this arrangement, the pump delivered 100 gpm at a 100-ft head. At shut-off, a head of 120 ft was obtained. Figure 76 is a photograph of this pump and the reservoir. QOrifice Meters Two thin-plate VDI-type orifices made by the Taylor Instrument Company having diameters of 1.530 and 0.765 in. were used alternately with a =134= UNCLASSIFIED PHOTO 27739 - Fig. 76. Centrifugal Pump ond System Reservoir -_— -135- 50-in. mercury-filled U-tube manometer connected across the orifice for meas- uring core flow rates. These diameters corresponded to l/2-pipe diameter and 1/4-pipe diameter, respectively. A measurable flow rate range of one to 100 gpm was obtainable. A 30-diameter length of straight pipe preceded the orifice position. Rotameters A Brooks rotameter with a measurement range of 0.5 to 6.0 gpm and a Fischer and Porter rotameter of range 2 to 20 gpm were installed at a later date than the flowmeters. Control Valves Three globe-type valves were connected in a parallel arrangement to give good flow rate control over the entire range. A 3-in. valve, a l-in. valve, and a L/E—in. valve were used. Temperature Control A calrod immersion heater and a cooling coil made of copper tubing with cold water running through placed in the reservoir served to control the water temperature. Core Test Stand The core models were assembled on a support which formed part of the re- circulating loop system, and which served to align the inner and outer shells. A quickly removable section of pipe which served as the inlet to the core models facilitated access for changing or performing any alterations on the core models. A.X}ew of the test stand with a core model in place on 1ts support with #;:\ S thé femovable inlet pipe in place was shown in Figure 11. T | -136- \ REFERENCES 1. Poppendiek, H. F., Palmer, L. D., "Forced Convection Heat Transfer in Pipes with Volume Heat Sources Within the Fluids," ORNL-1395, February 1953. 2. Poppendiek, H. F., Palmer, L. D., "Forced Convection Heat Transfer Be- tween Parallel Plates and in Annull with Volume Heat Sources Within the Fluids," ORNL-1701, May 195k. 3. Poppendiek, H. F., Palmer, L. D., "Application of Temperature Solutions for Forced Convection Systems with Volume Heat Sources to General Con- vection Problems," ORNL-1933, September 1955, 4, Bell, R. BE., "Investigation of the Fluid Flow Pattern in a Model of the 'Fireball' Reactor," ORNL Memo, Index No. Y-F15-11, Sept. 4, 1952, 5. Bradfute, J. 0., et al, "Reactor Hydrodynamics," ANP Quarterly Progress Report, ORNL-1729; p 102, June 10, 195h. 6. Whitmsn, G. D., Stelzman, W. J., Furgerson, W. T., Aircraft Reactor Engineering Division, ORNL, Personal communication of unpublished velocity and pressure data for the ART "21-inch" Core. 7. Buri, "Dissertation,” Zurich, 1931. 8. Nikuradse, Johann, "Untersuchungen Uber Die Stromungen Des Wassers In Konvergenten Und Divergenten Kenalen," Forschungsarbeiten, V.D.I., Volume 289, pp 1-49, 1929. 9. Keller, J. D., J. Appl. Mech. 16, pp 77-85 and p 320, (1949). 10. Bradfute, J. 0., "Qualitative Velocity Information Regarding the ART Core; Status Report IV," ORNL CF 54-12-110, December 14, 1954, 1l1. Bradfute, J. 0., Lynch, F. E., Muller, G. L., "Fluid Velocity Measured in the 18-Inch ART Core by a Particle-Photographic Technique," ORNL CF 55-6-137, June 21, 1955. 12, Muller, G. L, "Qualitative Estimates of the Velocity Profiles in the 21-Inch ART Core," ORNL CF 55-7-92, July 20, 1955, 13. Muller, G. L., Lynch F. E., "Qualitative Velocity Informastion Regarding Two Constant-Gap Core Models," ORNL CF 55-10-49, October 3, 1955. 1k, 15. 16. 17. 18. 19. 20. 21. 22. 23. 2k, 25. REFERENCES (Continued) Muller, G. L, Bradfute, J. 0., "Qualitative Velocity Profiles with Rotation in 18-Inch ART Core," ORNL CF 55-3-15, March 1, 1955. Lynch, F. E., and Muller, G. L., "Qualitative Velocity Profiles with &8 Rotational Component at the Inlet of the 2l1-Inch Core Model; G. F. Wislicenus Design," ORNL CF 55-10-50, October 3, 1955. Lynch, F, E., and Bradfute, J. 0., "Qualitative Velocity Profiles in 18-Inch ART Core with Increased Turbulence at its Inlet," ORNL CF 55-5-132, May 10, 1955. Muller, G. L., "Qualitative Velocity Informstion Regarding the 18-Inch ART Core with Turbulator Vane Set No. 1 in Entrance,"” ORNL CF 55-6-1T7k, June 27, 1955. Lynch, F. E., "Qualitative Velocity Information Regarding the 18-Inch ART Core with Vane Set No. 2 in Entrance," ORNL CF 55-6-173, June 27, 1955. Muller, G. L., and Lynch, F. E., "Qualitative Velocity Informstion Re- garding the Quarter-Scale Model of the 18-Inch ART Core and the 5/22-Scale Model of the 21-Inch ART Core with the Pratt and Whitney Swirl Nozzles at the Inlet," ORNL CF 55-10-48, October 3, 1955. Muller, G. L., Lynch, F. E., "Effects of Screen Packing in the ART 21-Inch Core and in an RMR Core Designed to Concentrate the Packing in the Core Entrance Region,” ORNL CF 56-12-5, December 20, 1956. Stelzmen, W. J., Whitman, G. D., Furgerson, W. T., Personal communica- tion of photographic data on dye filement fluctuations in full-size model of ART Core. Goldstein, S., Modern Developments in Fluid Dynamics, Corrected Impression, Oxford University Press, Amen House, London, E. C. 4, 1950, Vol 1, pp 58-59. Wattendorf, F. L., "A Study of the Effect of Curvature on Fully Devel- oped Turbulent Flow," Proc. Roy. Soc. A, Vol 148, 1935, pp 565-598. Baines, W. D., and Peterson, E. G., "An Investigation of Flow Through Screens," Trans. ASME, July 1951, pp 46T7-L480. Schubauer, G. B., and Spangenburg, W. G., "Effect of Screens in Wide- Angle Diffusers,” NACA Report 949, 1949, 4“)7 ' ‘: @ — -138- S 26. 27. 28. 29. 30. 31. 32. 33 34, 35. 36. 37 38. 39. REFERENCES (Continued) Stumpf, H. J., and Wilner, B. M., ANP Quarterly Progress Report, March 10, 1954, ORNL-1692, pp 37-39. Poppendiek, H. F., et al., Analytical and Experimental Studies of the Temperature Structure within the ART Core, ORNL-2198, Janusry 1957. Laufer, J., "Investigation of Turbulent Flow in a Two-Dimensional Channel," NACA Report 1053, 1951. Stelzman, W. J., Furgerson, W. T., ANP Quarterly Progress Report, September 10, 1956, ORNL-2157, Parts 1-5, p 28 (Figure 1.1.10). Goldstein, S., Modern Developments in Fluid Dynamics, Corrected Impression, Oxford University Press, Amen House, London, E. C. 4, 1950, Vols. 1 and 2, plate 12-c, opp. p 76; plate 17-a, opp. p 82; plate 33-d4, opp. p 552. Palmer L. D., Lynch F. E., Winn, G. M., Instantaneous Velocity Profile Measurement by Photography, ORNL-2257 (to be published). Mills, C. B., and Reese, H., "Design Study of an ANP Circulating-Fuel Reactor," WAD-1930, November 30, 1954. Nuclear Propulsion Program, Engineering Program Report No. 12, April 1 - June 30, 1954, PWAC-542, pp 63-6k, Nuclear Propulsion Program, Engineering Program Report No. 13, July 1 - September 30, 1954, PWAC-543, pp 63-6k, Nuclear Propulsion Program, Engineering Program Report No. 1k, October 1 - December 31, 1954, PWAC-54k, pp 52-53. Nuclear Propulsion Program, Engineering Program Report No. 15, January 1 - March 31, 1955, PWAC-551, pp 42-4kL. Nuclear Propulsion Program, Engineering Program Report No. 16, April 1 - June 30, 1955, PWAC-552, pp 45-46, Nuclear Propulsion Program, Engineering Program Report No. 17, July 1 - September 30, 1955, PWAC-553, pp 48-49. Nuclear Propulsion Program, Engineering Program Report No. 18, October 1 - December 31, 1955, PWAC-554, pp 56-58. - -139- 40, 4. 42, 43. Gk, REFERENCES (Continued) Nuclear Propulsion Program, Engineering Program Report No. January 1 - March 31, 1956, PWAC-561, p 66. Nuclear Propulsion Program, Engineering Program Report No. April 1 - June 30, 1956, PWAC-562, pp 62-63. Nuclear Propulsion Program, Engineering Program Report No. July 1 - September 30, 1956, PWAC-563, p 65. Nuclear Propulsion Program, Engineering Program Report No. October 1 - December 31, 1956, PWAC-564, p 6k, Nuclear Propulsion Program, Engineering Program Report No. January 1 - March 31, 1957, PWAC-565, pp 61-6k. 19, 20, 21, 22, 23, i . #’ - i ' % o1k ORNL-2199 \ C-84 - Reactors-Speaial Features of Aircraft Reactors M-3679 (20th ed. Rev.) INTERBAL DISTRIBUTION A 6!}, Ccenter T 50. E. S. Be@bis & i 30gy Library L >l. C. J. Dagron e, Physics Library ; 52. ~if§*\ -nkenship L g2l Research Library & 53« § W L MaBor Experimental 5 engWeering Library - BNy Records Departmengt - +oMk Records, ORNL R.C. einWarsg oo -25) 2) B. Emlet P. Murray A. Swartout . BE. H. Taylor D. Shipley . L. Nelson . S. C. Lind 2l. F. L. Culler 22, A H Snk 1 23 wier 25, 7} Cohen 26. K& . M. Copenhaver 27. C¥ W. R. Gall 28. R.ES. | '3y W. R. Gambill 29. A} S. Housgpholde Ble N. D. Greene 30. j Hafi 11 . Keyes 31. . WihtRrs Krekoviak 32, : da% Hoffman 33. D. Phlxl D & I Lunch 4. AEJV BiJ - N. Lyon 35. R.}A. G | Powers 36. J. @A Thomas 37. R.JR: Wantland 38. R. je.cPherson 39. ' Rzley MA . } W B.:McD¥ald #R. Vo Msgh blian = A. M.- Perry )0. A. W.' SavolaNgen 91. R. D. Schulthqes 92, W. L. Scott 93. W. J. Stelzman oy -1lk2- Douglas‘t 100. 101. 102. L. 103. E. & 104, ] ., 23Technlcal lerary, 105. H.;C, gL ngerence Sectlon g A 116.°D fon of RWlarch and De i L 117~ 1L9 Mrce BalliNgc Missilg " 120- lgg sBoelng, Sell y L § Boeing, Wic . 123. AFPR Curtlss-ergh Cl¥tj 12L. AFPR, Douglas, Long I@ech 125-127. AFE r"Bouglats, Santa ica 128. AFPE ‘iockheed Burbank 129-130. &FER, LOCkheed Marietta 131. AFER% E0r€h American, Cano@ 132, AAFPB,'NGfth American, Downsg 133-13k. Air)Force Specjpl. Weapons 135. Air Materiel Cfmimt : 136Q‘A1#-§esearch aff Ent Corfgd (RDGN) 137’.Air;Besearch .f fment Comm?® (RDTAPS) 138-151." Ppment Comme a . 152 pence Center 153-155. Dnvalr, Fort Wor® 156 k _xohs»Offlce ‘157 Argonne Natid © 158, Armed Forces . Sl fiWeapons Project, Sanifg - 159. Armed Forces . Weapons Project, Washi¥gton 160, Assistant : ' 161-165. Atomic Engli-8 167. Atomics Cefihs 168. Battel emi 169-170. BettigPlarjti(} 171. Bureg of i 172. 2 173. j ; 17k, % San Diego 175. f'cMartln, Baltimore 176 Alrcraft Bethpage 177 Bureau of Y ds and Docks ’ : 178. Chicago Op~fa#10ns Office ; 179. Chicago Patgn% Group / 180. Curtiss-Wri§ ’. * Corporation . B -143- G 2 181 Engineer Reseaf%h and Development Laboratories L = 1gg et . Lt o - / s T f% /‘ SUNIR A > A ps Scientific Laboratory ipory Committee for S ,heutics, Cleveian@‘ Committee fopd¥ronautics, Washington 192, Marquar® 193. Martin CO 194. National Ad 195. National Adv , 196. Naval Alr Deve ' 197 . RETHORE R, 198, Glisme Dot ........ 200 g emv«ep‘e?at o 201, (mclenr IEVETEERent 202, Nuclear Metals, Ing 203. Office of Naval B€arch 20k, Office of the C}Ff of Naval 205. Patent Branchshington 206-227. Pratt and W ey Aircraft Divis :tion of Americsa W ations (OP-361) s (1 copy each to G. Beardsley, C. C. Bige " A. I. Chalfant, W. SQ@liss, R. A. Doak, E. R. Dytko, H. C. Gra . Hoenig, C. E. HoltsinA K. J. Kelley, R. W. Kelley, A. Y.,Q fhan, B. A. Schmickrath, R. 1N Rt rough, R. N. Wallace, D Pfl,, N LT 2 ., N o _ 228. San K 229. Sangl@ Corpora - 230. 8§ Bl of AV1ation Medicine 231. g ania-Corning Nuclear Corporation Pchnical Research Group USAF Headquarters USAFT Project Rand . U. S, N&Evg¥ Radiologicel Defense L&boratory Unlversifiyh§@30fiiifernla Radiation laboratory, Livermore Wright Afi?@fififiéfiofiment Cénter (WE0SI-3) ‘ SRR o Sérvice Exten51on, Oak Ridge PR x" R RN R T