Jtam.dvi JOURNAL OF THEORETICAL AND APPLIED MECHANICS 45, 3, pp. 557-568, Warsaw 2007 FLOW PATTERNS GENERATED BY A STRONG MAGNETIC FIELD Elżbieta Fornalik AGH, University of Science and Technology, Faculty of Non-Ferrous Metals, Poland e-mail: elaf@agh.edu.pl The influence of a strong magnetic field on a paramagnetic fluid in a cy- lindrical enclosure of thermosyphon-like geometry was studied. A variety of spoke patterns was experimentally obtained and recorded. The analysis led to some data about the temperature field and, indirectly, about the flow structure. The results related to the heat transfer rate confirmed the conc- lusions coming from the flow visualization. Key words: spoke pattern, thermosyphon, magnetic field 1. Introduction In nature, a variety of different kinds of patterns can be found. The beauty of nature can also be recognized in industrial processes, for example in solidifica- tion (dendrites) or spoke patterns during growth of a single crystal. The spoke pattern is also observed in a thermosyphon-like enclosure in which natural convection appears. Pattern formation for natural convection has been studied since the be- ginning of the 20th century. Benard (1901) observed appearance of hexagonal cells as a result of instabilities in a shallow layer of a fluid heated from below. Rayleigh (1916) presented the theory related to Benard’s experiment. He pro- posedanon-dimensional parameter called theRayleighnumber to characterize the convection. Theworks donebyJeffreys (1928, 1930), Pellew andSouthwell (1940) or Chandrasekhar (1961) should also be mentioned. Recently, Hof et al. (1999) reported experimental results of natural convection investigations in the vertical cylinder heated from below. They studied the stability of flow and found that at the same Rayleigh number various patterns can exist. The onset of convective instabilities in cylindrical cavities heated from below was described by Touihri et al. (1999). They numerically analyzed this phenome- non with the attention paid to the nonlinear evaluation of convection. The 558 E. Fornalik multiple steady states for the sameRayleigh numberwere obtained byBoron- ska and Tuckerman (2004) who simulated numerically Hof’s experiment (Hof et al., 1999). Numerous studies on the dynamic patterns of convection in a melt have been carried out, and many researchers studying crystal growth have repor- ted the occurrence of ”spoke patterns” on the melt surface. The flow loses its stability and the axisymmetric structures break its symmetry and become non-axisymmetrical. It is observed that flow is divided into several identical rolls. Adjacent roll structures (sectors) have an opposite azimuthal veloci- ty component. Moreover, computed results show that the azimuthal velocity of the structure modifies the temperature profile from axisymmetric to non- axisymmetric. Numerical computation applied to the analysis of buoyancy driven flow in a vertical cylinder shows a non-axisymmetric character of the flow even for axisymmetric boundary conditions. Kowalewski et al. (1998) and Gelfgat et al. (1999) observed the same spoke structures and axisymmetry breaking instability with a large azimuthal number. The pattern formation in a cylindrical enclosure of thermosyphon-like geo- metry was described, for example, by Japikse et al. (1971), Mallinson et al. (1981) or Ishihara et al. (2002a,b). They observed appearance of alternate streams of a hot and cold fluid in various numbers depending on the fluid properties, geometrical and thermal conditions. All mentioned examples were related to the natural convection phenome- non. However, due to the development of superconducting magnets, a new phenomenon called ”magnetic convection” could be investigated. Braithwaite et al. (1991) as the first researcher reported the influence of magnetic field on the natural convection of a paramagnetic fluid. Their work related to thema- gnetic field was continued, among others, by Ikezoe et al. (1998), Wakayama (1993), Wakayama and Wakayama (2000), Kaneda et al. (2002), who studied the influence of magnetic field on many other phenomena in various research areas.Wide knowledge of the magnetic field effect on engineering processes is described in a book by Ozoe (2005). Thepresentpaper gives an insight into themagnetically inducedflow in the thermosyphon-like cylinder with the attention paid to the pattern formation. It should be added that in such a configuration of the cylinder, the natural convection is very limited and no pattern can be recognized. 2. Basics Ordinary substances (like water or air) show somemagnetic effects, although very small. This smallmagnetism is of two kinds. Somematerials are attracted Flow patterns generated by a strong magnetic field 559 toward the magnetic field, others are repelled (Feynman et al., 1977). The substances, which are repelled, are called diamagnetic, whereas the substances which are attracted are called paramagnetic. The paramagnetic materials have positive magnetic susceptibility which is usually a function of absolute temperature in contradiction to diamagnetic materials whose magnetic susceptibility is negative and independent of tem- perature. The equation taking into account Curie’s law and describing the magnetic force can bewritten in a form obtained byTagawa andOzoe (2002) Fm = −ρ0χ0β(θ−θ0) µm ∇b 2 (2.1) where ρ0 denotes reference density at the reference temperature θ0, χ0 – re- ferencemagnetic susceptibility, β – thermal expansion coefficient, θ – tempe- rature, θ0 – reference temperature, µm –magnetic permeability, b –magnetic induction. The driving force for convection is usually the density difference between hot and cold regions of the fluid. If the fluid has magnetic susceptibility that varies with temperature, themagnetic forces, rather than buoyancy, can drive convective motion (Braithwaite et al., 1991). The magnetic buoyancy force which drives convective motion is proportional to the gradient of square of magnetic induction. 3. Experimental enclosure The experimental enclosure is presented in Fig.1. The cylinder diameter was 0.04m and the heated and cooled side walls were made of copper of 0.028m height separated by thin Plexiglas cylindrical plate of 0.004m thickness. Four holes were drilled from the side wall of this plate to place four Light Emitting Diodes (LEDs) for visualization of the temperature field in the chosen cross- section. The top and bottom end plates of the cylinder weremade of Plexiglas of 0.005m thickness. The outside surface of the enclosure side wall was heated with a rubber-coated nichrome wire. The wire was connected to a DC power supply whose current and voltage were constantly controlled by multimeters (Fig.2). The sidewall of the second half of the cylinderwas kept at a constant temperature with water from a constant temperature bath. The temperature difference between the heated and cooled parts of the enclosure wasmeasured by Copper-Constantan thermocouples. Two were located on the outer side of the heated wall, while the other two were attached to the coolant outlet and inlet. The thermocouples were connected to a data acquisition system. The measured temperature was recorded and stored for further analysis. 560 E. Fornalik Fig. 1. Experimental enclosure or apparatus Fig. 2. Scheme of the experimental set-up 4. Experimental fluid A 50% volume glycerol aqueous solution was chosen as the working fluid, because glycerol itself is hard to handle. Thismixture is diamagnetic however, the aim was to study a paramagnetic fluid. Therefore, crystals of gadolinium nitrate hexahydrate (Gd(NO3)3 × 6H2O) were added to the fluid increasing its magnetic susceptibility up to χexp =13.926 ·10 −8m3/kg. The molar concentration of gadolinium nitrate hexahydrate was 0.5mol/kg. Before starting the main part of the experiment, the density, ma- gnetic susceptibility and viscosity of the fluidweremeasured.Other properties were estimated from [26] for the 50% volume glycerol aqueous solution. The properties of the working fluid are listed in Table 1. The liquid crystal slurry Flow patterns generated by a strong magnetic field 561 was mixed with the working fluid to visualize the temperature field in the middle-height cross-section. Table 1. Properties of 50% aqueous solution of glycerol at θ0 = 298K (the properties marked by asterisk were estimated from [26]) Property Value Unit α∗ 1.1415 ·10−7 m2/s β∗ 0.445 ·10−3 K−1 λ∗ 0.422 W/(m·K) µexp 6.145 ·10 − ±0.064 ·10−3 Pa·s νexp 4.80 ·10 −6 ±0.05 ·10−6 m2/s ρexp 1281±1 kg/m 3 χexp 13.926 ·10 −8 ±0.329 ·10−8 m3/kg 5. Parameters The experiment was carried out for various Rayleigh numbers defined as Ra= gβ(θhot−θcold)r 3 0 αν (5.1) where g is the gravitational acceleration, β – thermal expansion coefficient, θhot – temperature of the heated part of the side wall, θcold – temperature of the cooled part of the side wall, α – thermal diffusivity of the fluid, ν – kinematic viscosity, r0 – inner radius of the cylinder. Thermalmeasurements were carried out to investigate the influence of the magnetic field on the heat transfer rate. The Nusselt number was defined as follows Nu= Qnet conv Qnet cond (5.2) The net convection (Qnet conv) and net conduction (Qnet cond) heat fluxes were estimated by the method invented by Ozoe and Churchill (1973) and applied, e.g., inMaki et al. (2002) Qnet cond =Qcond−Qloss (5.3) Qnet conv =Qconv−Qloss The net convection heat flux (Qnet conv), Eq. (5.3)1 was estimated as the difference between the total heat supply during the convection experiment 562 E. Fornalik and the heat loss. The net conduction heat flux (Qnet cond), Eq. (5.3)2 was estimated as the difference between the total heat supply during the conduc- tion experiment and the heat loss. The ratio of these net heat fluxes gives the Nusselt number. The convection heat flux (Qconv) was calculated from the electrical current and voltage values, while the conduction heat flux (Qcond) from an additional conduction experiment described in detail in Fornalik et al. (2005). The heat loss was assumed to depend on the heater temperature itself and not on themode of heat transfer inside the enclosure. 6. Procedure The experimental enclosure filled with the experimental fluid and insulated with a vinyl foil was placed into the bore of a superconducting magnet. The middle height cross-section of the cylinder was placed at 0.01m from the ope- ning level of the bore. That position guaranteed the minimum radial compo- nent of the magnetic buoyancy force in the tested volume. The experimentwas carried out formagnetic induction from 0 to 5Twith a step of 1T. The visualization of fluid temperature was done by the encap- sulated liquid crystal slurry. Depending on the temperature of the fluid, the liquid crystals showed different colors: the blue color represented hot fluid, while the red color indicated the cold fluid. The temperature indication range for color visualization was 291-294K. The color images of flow modes were taken by a digital camera working in a long exposure time of 4 seconds. The thermal measurements by thermocouples were carried out to investigate the influence of the magnetic field on the heat transfer rate. The procedure and definitions were reported in detail in Fornalik et al. (2005). The experimental apparatus was maintained at a constant environmental temperature. At the beginning, the cooling water bath and the heater power supply were switched on. The temperature level and the value of supplied power were set. The temperatures on the heated and cooled side walls were monitored.When the system reached the steady state after 90-120minutes (it depended on the temperature difference), temperatures were recorded and, at the same time, thepicture of fluid temperaturefieldwas taken.The steady sta- tewas assumedwhen the temperature stayed at the same level for 30minutes. Then, the magnetic field was applied to the system. It should be emphasized that after each step (for example after changing the magnetic induction from 1T to 2T), the system was kept in a constant condition to obtain another steady state.When it was reached (after about 40-60minutes – depending on the magnetic induction), the temperature values and the fluid flow structure were recorded. Flow patterns generated by a strong magnetic field 563 7. Results Figure 3 shows the visualized isotherms for selectedRayleigh numbers and va- rious strengths ofmagnetic induction. The pictures in Fig.3 at 0Twere taken for various Rayleigh numbers without the magnetic field. The uniform color of the pictures indicates the presence of an isothermal layer of the fluid in the middle height cross-section. The color represents the mean fluid temperature. The spoke pattern was not observed, which suggests that convective motion was not present. Fig. 3. Isotherms for various strengths of the magnetic field in the stable configuration, (a) Ra=0.14 ·105, (b) Ra=2.79 ·105, (c) Ra=5.22 ·105 When the level of 2T was achieved, the spokes appeared suddenly in the visualized cross-section. In Fig.3a at Ra = 0.14 · 105 and 2T, three spokes suddenly appeared, while in Fig.3b at Ra = 2.79 · 105 seven spokes and in Fig. 3(c) at Ra = 5.22 ·105 nine spokes. The number of spokes increased in Fig.3a from 0 to 8, In Fig.3b from 0 to 14 and in Fig.3c from 0 to 17 with an increase in the magnetic induction from 0T to 5T. The appearance of spoke patterns is a sign of induced convection in the system. Increasing with the magnetic induction, the number of spokes is related to enhanced convection. Figure 4 presents calculated isotherms at the middle-height cross-section with the three-dimensional demonstration of long time streak lines for better understanding of spokes origins. The complete numerical analysis of the phenomena is presented in Filar et al. (2006). The fluid moves upwards near the lower heated side wall, then proceeds toward the enclosure center at themiddle-height, andnear the center ascends toward the top of cylinder. Finally, the fluid descends along the upper 564 E. Fornalik Fig. 4. Computed isotherms at the cylinder mid-height and three-dimensional demonstration of long time streak lines for the stable system at Ra=1.89 ·105 and b=2T cooled side wall. This trajectory is typical for thermosyphon systems andwas reported by Japikse et al. (1971), Ishihara et al. (2002a,b). In Fig.5, the configuration of gravitational buoyancy (Fg) and magnetic buoyancy (Fm) forces in the experimental enclosure are presented. This sche- matic drawing can help one to understand the fluid behaviour. Fig. 5. Configurations of the system (Fg – gravitational buoyancy force, Fm –magnetic buoyancy force) The relatively cold fluid inside the lower cooled region having a higher den- sity tended to stay in the lower part due to the gravitational buoyancy force. However the magnetic buoyancy force acted upward on the colder fluid with a larger value of the magnetic susceptibility. On the other hand, the hot fluid was kept in the upper part of the cylinder by the gravitational buoyancy force, but it was also repelled by the magnetic buoyancy force due to the smaller Flow patterns generated by a strong magnetic field 565 value of themagnetic susceptibility. Therefore, without themagnetic field, the systemwas almost thermally stable showing no pattern. It was thermally sta- ble until themagnetic buoyancy forcewas smaller or equal to the gravitational buoyancy force.When themagnetic buoyancy force acting in the opposite di- rection exceeded the gravitational buoyancy force, the magnetic convection was induced in the system. It could be deduced from the appearance of spoke patterns characteristic for natural convection in a thermosyphon-like enclosu- re with the upper side wall cooled and lower one heated. The increasing of themagnetic induction caused an increase in the number of spokes suggesting enhanced convective motion. Figure 6 shows the Nusselt number versus the maximum magnetic in- duction. The measurements were done for various Rayleigh numbers. For the maximummagnetic induction of 0T, the Nusselt number was close to one for all cases except Ra=5.22 ·105. It wasmentioned above that without thema- gnetic field the system is ”almost” thermally stable, however slow convective motion is present in the vicinity of side walls. As it can be seen, for high Ray- leighs numbers, the convection close to the wall is ”slower” because the heat transfer rate is close to 6. For the magnetic induction of 2T, the suppression of convective motion can be suggested by the decreasingNusselt number. The magnetic buoyancy force compensates the gravitational buoyancy force. Fur- ther growth of the magnetic induction causes growth of the Nusselt number, which means intensified heat transfer by enhancedmagnetic convection. Fig. 6. The Nusselt number versus the maximummagnetic induction for various Rayleigh numbers 566 E. Fornalik 8. Conclusions Experimental visualizations of the flow and heat transfer rates were presented for a single-phase thermosyphon under various thermal and magnetic condi- tions.Without amagnetic field, the convective flowwas not observed and the Rayleigh number hadno visible influence on the flow structure.After applying amagnetic field, the convective flow was obtained. It was found that the ma- gnetic buoyancy force of about 1T appeared to be equal to the gravitational buoyancy force. Magnetic fields greater than 1T induced the convective flow. In the discussed experiment, the induction of convective flow resulted in the appearance of a spoke pattern in the middle height cross-section and also in an increase of the Nusselt number. Acknowledgement A part of this work was supported under grant No. 10.10.180.345 sponsored by AGH, University of Science and Technology. References 1. Benard H., 1901, Les tourbillons cellulaires dans une nappe liquide transpor- tant de la chaleur par convection en regime permanent,Annales de Chiemie et de Physique, 23, 62-144 2. BoronskaK., TuckermanL.S., 2004,Multiplicity of patterns in cylindrical convection,Mechanics of the 21st Century, edit. byGutkowskiW.,Kowalewski T.A., Springer 3. Braithwaite D., Beaugnon E., Tournier R., 1991,Magnetically control- led convection in a paramagnetic fluid,Nature, 354, 134-136 4. Chandrasekhar S., 1961,Hydrodynamic andHydromagnetic Stability, Dover Publications, NewYork 5. FeynmanR.P., LeightonR.B., SandsM., 1977,The Feynman Lectures on Physics, Addison-Wesley Publishing Company 6. FilarP.,FornalikE.,TagawaT.,OzoeH., SzmydJ.S., 2006,Numerical and experimental analyses of magnetic convection of paramagnetic fluid in a cylinder, J. Heat Transfer, 128, 183-191 7. Fornalik E., Filar P., Tagawa T., Ozoe H., Szmyd J.S., 2005, Experi- mental study on the magnetic convection in a vertical cylinder, Experimental Thermal and Fluid Science, 29, 971-980 8. Gelfgat A.L., Bar-Yoseph P.Z., Solan A., Kowalewski T.A., 1999, An axisymmetry-breaking instability of axially symmetric natural convection, J. of Transport Phenomena, 1, 173-190 Flow patterns generated by a strong magnetic field 567 9. Hof B., Lucas P.G.J., Mullin T., 1999, Flow state multiplicity in convec- tion,Physics of Fluid, 11, 2815-2817 10. Ikezoe Y., Hirota N., Nakagawa J., Kitazawa K., 1998, Making water levitate,Nature, 393, 749-750 11. Ishihara I., Fukui T., Matsumoto R., 2002a, Natural convection in a ver- tical rectangular enclosure with symmetrically localized heating and cooling zones, Int. J. of Heat and Fluid Flow, 23, 366-372 12. Ishihara I., Imanishi R., Fujiwara M., Matsumoto R., 2002b, Natural convection in a single-phase closed thermosyphon, Proceeding of the 10th Int. Symposium on Flow Visualization, Kyoto 13. Japikse D., Jallouk P.A., Winter E.R.F., 1971, Single-phase transport processes in the closed thermosyphon, Int. J. of Heat and Mass Transfer, 14, 869-887 14. Jeffreys H., 1928, Some cases of instability in fluid motion, Proceedings of the Royal Society, 118 15. Jeffreys H., 1930, The instability of a compressible fluid heated below,Pro- ceedings of the Royal Society, 26 16. Kaneda M., Tagawa T., Ozoe H., 2002, Convection induced by a cusp- shaped magnetic field for air in a cube heated from above and cooled from below, J. of Heat Transfer, 124, 17-25 17. Kowalewski T.A., Cybulski A., Rebow M., 1998, Particle image veloci- metry and thermometry in freezing water, CD-ROM Proceedings of 8th Int. Symposium on Flow Visualization, Edinburgh, 47.1-47.11 18. Maki S., Tagawa T., Ozoe H., 2002, Enhanced convection or quasi- conduction states measured in a super conducting magnet for air in a vertical cylindrical enclosure heated from below and cooled from above in a gravity field, J. of Heat Transfer, 124, 667-673 19. Mallinson G.D., Graham A.D., Vahl Davis G., 1981, Three-dimensional flow in a closed thermosyphon, J. of Fluid Mechanics, 109, 259-275 20. Ozoe H., 2005,Magnetic Convection, Imperial College Press 21. Ozoe H., Churchill S.W., 1973, Hydrodynamic stability and natural co- nvection in newtonian and non-Newtonian fluids heated from below, AIChE Symposium Series, Heat Transfer, 69, 126-133 22. Pellew A., Southwell R.V., 1940, On maintained convective motions in a fluid heated from below, Proceedings of the Royal Society of London, 173, 312-343 23. Rayleigh L., 1916,On convection currents in a horizontal layer of fluid, when the higher temperature is in the under slide,Philosophical Magazine, 32 24. Tagawa T., Ozoe H., 2002, Convective and diffusive phenomena of air in a vertical cylinder under strongmagnetic field,Num. Heat Trans., 41, 1-14 568 E. Fornalik 25. Touihri R., Ben Hadid H., Henry D., 1999, On the onset of convective instabilities in cylindrical cavities heated from below. I. Pure thermal case, Physics of Fluids, 11, 2078-2088 26. VDI-Wärmeatlas, VDI-Verlag, 1997 27. Wakayama M., WakayamaN.I., 2000,Magnetic acceleration of inhaled and exhaled flows in breathing, Japanese J. of Applied Physics, 39, 262-264 28. Wakayama N.I., 1993,Magnetic promotion of combustion in diffusion flames, Combustion and Flame, 93, 207-214 Wpływ silnego pola magnetycznego na strukturę przepływu Streszczenie Przedstawione zostały badania dotyczące struktury przepływuw geometrii cylin- drycznej typu odwrócony termosyfon (tzn. górna część cylindra była grzana, a dolna chłodzona). Naczynie eksperymentalne napełnione cieczą paramagnetyczną zostało umieszczonew przestrzeni badawczejmagnesu nadprzewodzącego, generującego silne pole magnetyczne (maksymalna indukcja magnetyczna 5 Tesli). Badano wpływ po- la magnetycznego na strukturę przepływu, dokonano wizualizacji pola temperatury przy pomocy ciekłych kryształów oraz pomiaru temperatury za pomocą termopar. Bez pola magnetycznego oraz dla indukcji magnetycznej poniżej 1 Tesli w naczyniu nie obserwowano ruchu konwekcyjnego, a co za tym idzie również żadnej struktury przepływu.Dla indukcjimagnetycznej powyżej 1Tesli w naczyniu eksperymentalnym została zapoczątkowana konwekcjamagnetyczna, co objawiło się pojawieniem struk- tury szprychowej. Struktura szprychowa, typowa dla konwekcji naturalnej w termo- syfonie (tzn. górna część cylindra chłodzona, a dolna grzana) uległa zmianie – wraz ze wzrostem indukcji magnetycznej wzrastała liczba płatków, co sugerowało intensy- fikację konwekcji.Wyniki wizualizacji zostały potwierdzone przez pomiary termiczne oraz wyznaczoną liczbę Nusselta. Manuscript received January 23, 2007; accepted for print April 13, 2007