1: %\documentclass[12pt,preprint]{aastex}

2: %\documentclass[manuscript]{aastex}

3: \documentclass[preprint2]{aastex}

4: 

5: \shorttitle{Traveling wave MRI}

6: \shortauthors{R\"udiger et al.}

7: 

8: \begin{document}

9: 

10: \title{The traveling wave MRI in cylindrical Taylor-Couette flow:

11:         comparing wavelengths and speeds in theory and experiment}

12: 

13: \author{G\"unther R\"udiger}

14: \affil{Astrophysikalisches Institut Potsdam, An der Sternwarte 16,

15: D-14482 Potsdam, Germany\\gruediger@aip.de}

16: 

17: \author{Rainer Hollerbach}

18: \affil{Department of Applied Mathematics, University of Leeds,

19: Leeds, LS2 9JT, United Kingdom\\rh@maths.leeds.ac.uk}

20: 

21: \author{Frank Stefani, Thomas Gundrum, Gunter Gerbeth}

22: \affil{Forschungszentrum Rossendorf, P.O. Box 510119, D-01314 Dresden,

23: Germany\\F.Stefani@fz-rossendorf.de, Th.Gundrum@fz-rossendorf.de,

24: G.Gerbeth@fz-rossendorf.de}

25: 

26: \author{Robert Rosner}

27: \affil{Department of Astronomy and Astrophysics, University of Chicago,

28: Chicago, IL 60637, USA\\rrosner@astro.uchicago.edu}

29: 

30: \begin{abstract}

31: We study experimentally the flow of a liquid metal confined between

32: differentially rotating cylinders, in the presence of externally

33: imposed axial and azimuthal magnetic fields.  For increasingly large

34: azimuthal fields a wave-like disturbance arises, traveling along

35: the axis of the cylinders.  The wavelengths and speeds of these

36: structures, as well as the field strengths and rotation rates at

37: which they arise, are broadly consistent with theoretical predictions

38: of such a traveling wave magnetorotational instability.

39: \end{abstract}

40: 

41: \keywords{instabilities, magnetohydrodynamics, waves}

42: 

43: \section{Introduction}

44: 

45: The magnetorotational instability (MRI) arises in a broad range of

46: astrophysical problems, most importantly in accretion disks, where

47: it is generally accepted to be the source of the turbulence needed

48: for material to spiral inward and accrete onto the central

49: object \citep{BH91}.  Because of this crucial role that it plays in

50: astrophysics, there is considerable interest in trying to study the

51: MRI in the laboratory \citep{RRB04}.  One recent suggestion

52: \citep{HR05,RHS05} involves externally imposing combined axial and

53: azimuthal magnetic fields, which yields a new type of traveling

54: wave MRI.  In this letter we present experimental evidence of these

55: traveling waves, in accordance with the theory.

56: 

57: The MRI is a mechanism whereby a differential rotation flow that

58: satisfies the Rayleigh criterion, and is therefore hydrodynamically

59: stable, may nevertheless be magnetohydrodynamically unstable.  The

60: addition of a magnetic field allows angular momentum to be transferred

61: outward by the magnetic tension in the field lines, thereby bypassing

62: the Rayleigh criterion, which relies on individual fluid parcels

63: conserving their angular momentum.  The MRI is particularly relevant

64: to Keplerian flows such as those found in accretion disks, where

65: $\Omega\sim r^{-3/2}$, which would indeed be stable according to the

66: Rayleigh criterion.

67: 

68: The simplest design for attempting to reproduce the MRI

69: in the lab is based on the familiar Taylor-Couette problem,

70: consisting of the flow between differentially rotating cylinders.

71: While one cannot achieve precisely a Keplerian flow profile in this

72: problem, by appropriately choosing the rotation rates of the inner and

73: outer cylinders one can easily achieve a Rayleigh-stable profile,

74: which is all that is really required.  By suitably adjusting the

75: strength of an externally imposed magnetic field one should then be

76: able to destabilize the flow again, via the MRI.

77: 

78: Unfortunately, the situation is not quite so simple after all.  If

79: the imposed field is purely axial \citep{RZ01,JGK01}, the relevant

80: parameter for the onset of the MRI turns out to be the magnetic Reynolds

81: number $Rm=\Omega_{\rm i} r_{\rm i}^2/\eta$, which must exceed $O(10)$.

82: The hydrodynamic Reynolds number $Re=\Omega_{\rm i} r_{\rm i}^2/\nu$ then

83: exceeds $O(10^6)$, due to the extremely small magnetic Prandtl numbers

84: $Pm=\nu/\eta$ of liquid metals ($\nu$ is the viscosity, $\eta$ the

85: magnetic diffusivity).  Such large rotation rates can be achieved,

86: but for increasingly large Reynolds numbers end-effects become

87: increasingly important, and at $Re\gtrsim O(10^6)$ may well disrupt

88: the experiment \citep{HF04}.

89: 

90: In contrast, if a combined axial and azimuthal field is imposed

91: \citep{HR05,RHS05}, the relevant parameter turns out to be $Re$,

92: which must only be as large as $O(10^3)$ to obtain the MRI.  These

93: end-effects are therefore less severe.  The solutions in this case are

94: also somewhat different from those for purely axial imposed fields; one

95: obtains much the same Taylor vortices as before, but the whole pattern

96: now drifts along the length of the cylinders \citep{Knobloch}.  This

97: unfortunately introduces significant end-effects of its own, familiar in

98: other contexts, such as drifting dynamo waves \citep{TPK98}.  Nevertheless,

99: we will see that the experimental results in a bounded cylinder agree

100: reasonably well with the theoretical results in an unbounded cylinder.

101: 

102: \section{Experimental Results}

103: 

104: The experimental apparatus consists of a cylindrical annulus made of

105: copper, with $r_{\rm i}=4$ cm, $r_{\rm o}=8$ cm, and height 40 cm.  The top

106: endplate is made of plexiglass, and is stationary; the bottom endplate is

107: made of copper, and rotates with the outer cylinder.  An axial magnetic

108: field is imposed by running a current, up to 200 A, through a series of

109: coils surrounding the entire apparatus; an azimuthal magnetic field is

110: imposed by running a current, up to 8000 A, through a rod along the central

111: axis.  Field strengths of several hundred G can then be achieved, for both

112: $B_z$ and $B_\phi$.  The fluid contained within the vessel is a GaInSn alloy,

113: having density $\rho=6.4\ \rm g/cm^3$, viscosity $\nu=3.4\cdot 10^{-3}\ \rm

114: cm^2/s$ and magnetic diffusivity $\eta=2.4\cdot10^3\ \rm cm^2/s$, so $Pm=1.4

115: \cdot10^{-6}$.  Measurements were made by two ultrasonic transducers mounted

116: on opposite sides of the top endplate, 1.5 cm from the outer wall.  These

117: provided measurements of $v_z$ at these particular locations, along the

118: entire 40 cm depth of the container.  See also \citet{SGGRSSH} for further

119: details of the experimental setup.

120: 

121: For the results presented here, the rotation rates of the inner and

122: outer cylinders were fixed at $\Omega_{\rm i}=0.377\ \rm s^{-1}$ and

123: $\Omega_{\rm o}=0.102\ \rm s^{-1}$, so $\Omega_{\rm o}/\Omega_{\rm i}

124: =0.27$, within the Rayleigh-stable regime $\Omega_{\rm o}/\Omega_{\rm i}>

125: (r_{\rm i}/r_{\rm o})^2=0.25$.  The Reynolds number $Re=\Omega_{\rm i}

126: r_{\rm i}^2/\nu=1775$.  The axial field $B_z$ was also fixed, at 77.2 G,

127: corresponding to a Hartmann number $Ha=B_z r_{\rm i}/\sqrt{\rho\mu\eta\nu}

128: =12$.  The azimuthal field $B_\phi$ was then varied, between 0 and 350 G

129: (at $r_{\rm i}$), corresponding to electric currents up to 7000 A along the

130: central rod.

131: 

132: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

133: 

134: \begin{figure}

135: \epsscale{0.95}

136: \plotone{f1.eps}

137: \caption{Ultrasound measurements of the axial velocity $v_z$, with the

138: time-average removed.  Time is plotted on the horizontal axis; height

139: along the cylinder on the vertical axis.  The current along the central

140: rod varies from 0 to 7 kA as indicated, corresponding to

141: $B_\phi(r_{\rm i})$ up to 350 G.}

142: \end{figure}

143: 

144: \begin{figure}

145: \epsscale{0.95}

146: \plotone{f2.eps}

147: \caption{Contour plot of $Re_c$ as a function of $B_z$ and $B_\phi$.  On

148: the horizontal axes $B_z$ is given both in G, and in nondimensional form,

149: as $Ha=0.1555\cdot B_z/\rm G$.  On the vertical axes $B_\phi(r_{\rm i})$

150: is given both in G, and in terms of the required axial current $I_z$.  The

151: numbers beside individual contour lines denote values from 1000 to 6000;

152: the dashed contour line is 1775, the value used in the experiment.  The

153: five asterisks correspond to the five experimental runs in Fig.\ 1.}

154: \end{figure}

155: 

156: \begin{figure}

157: \epsscale{0.95}

158: \plotone{f3.eps}

159: \caption{The left panel shows the wavelengths, in cm, the right panel

160: shows the speeds, in mm/s, of the solutions at $Re=1775$.  As in Fig.\ 2, the

161: dashed curves are the stability boundary at this particular $Re$.  The three

162: asterisks correspond to the three experimental runs at 3, 5 and 7 kA,

163: where once again the wavelengths were approximately 8, 6 and 6 cm,

164: respectively, and the speeds 0.6, 0.8 and 0.8 mm/s.}

165: \end{figure}

166: 

167: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

168: 

169: Figure 1 shows the temporal variation of the axial velocity $v_z$, as a

170: function of $z$ along the container.  The (depth-dependent) time-average

171: has been subtracted, to remove the two Ekman vortices induced by the

172: endplates \citep{KJG04,HF04}.  As $B_\phi$ is increased, we see the gradual

173: emergence of ever more coherent structures, drifting at speeds of 0.5--1

174: mm/s.  Focussing attention on $z$ between 10 and 20 cm, where the waves are

175: most clearly defined, we obtain around 0.6 mm/s for 3 kA, and 0.8 mm/s for

176: 5 and 7 kA.  The corresponding wavelengths are 7--8 cm for 3 kA, and 5-6 cm

177: for 5 and 7 kA.  We claim that these structures are precisely the expected

178: traveling wave MRI.

179: 

180: Note though that end-effects do indeed play an important role in Fig.\ 1,

181: for example at the upper boundary, where the waves die away some 5--10 cm

182: from the end.  Another indication of the importance of end-effects,

183: and in particular the asymmetry between the two ends, can be seen in the

184: 0 A results: the asymmetry between top and bottom that is already visible

185: even in this case must be caused entirely by the endplates, as a purely

186: axial field does not distinguish between $\pm z$.

187: 

188: \section{Theoretical Analysis}

189: 

190: Figure 2 shows the critical Reynolds number for the onset of this traveling

191: wave MRI in an unbounded cylinder, as a function of the externally imposed

192: fields $B_z$ and $B_\phi$; these results were computed as in \citet{HR05} or

193: \citet{RHS05}, with conducting boundary conditions.  Provided $B_z>30$ G and

194: $B_\phi\gtrsim150$ G, Reynolds numbers of $O(10^3)$ are already sufficient.

195: If $B_\phi$ is less than 150 G, $Re_c$ gradually rises, until for $B_\phi=0$

196: we would have $Re_c>O(10^6)$, the familiar result from the analysis of purely

197: axial fields.  In contrast, if $B_z$ is less than 30 G, the instability

198: simply ceases to exist; that is, $Re_c$ has an essentially vertical asymptote

199: at this boundary.

200: 

201: The five asterisks in Fig.\ 2 correspond to the five plots shown in Fig.\ 1,

202: and the dotted contour line to the experimental value $Re=1775$.  We see

203: therefore that the experimental runs with $I_z=0$, 1 and 3 kA should be

204: stable, whereas the ones with 5 and 7 kA should be unstable.  This is

205: broadly in agreement with Fig.\ 1, although there even the supposedly stable

206: runs already show hints of traveling wave disturbances, particularly at 3

207: kA.  However, the waves are much more strongly developed for the 5 and 7 kA

208: runs, as predicted by Fig.\ 2.

209: 

210: Fig.\ 3 shows the wavelengths and speeds of the unstable solutions at $Re=

211: 1775$.  Wavelengths are in the range 10--20 cm, speeds around 1.1 mm/s (for

212: $B_z=77.2$ G).  The speeds agree rather well with Fig.\ 1; the experimental

213: result that 5 and 7 kA yield virtually the same speed, 0.8 mm/s, also

214: nicely matches the theoretical prediction that the speed should be almost

215: independent of $B_\phi$ (provided $B_\phi$ is large enough to be in the

216: unstable regime at all).  The wavelengths do not agree quite so well; the

217: values in Fig.\ 1 are barely half those in Fig.\ 3.  Presumably this is

218: again due to end-effects; a 20 cm long wave certainly could not traverse a

219: 40 cm long container without experiencing significant end-effects.

220: 

221: \section{Conclusion}

222: 

223: In this letter we have presented experimental evidence for the existence of

224: traveling wave disturbances in a liquid metal Taylor-Couette apparatus, and

225: shown them to be in reasonable agreement with the theoretical predictions,

226: particularly with regard to the wave speeds.  Future experimental work will

227: more thoroughly map out the entire $B_z$, $B_\phi$, $\Omega_{\rm i}$ and

228: $\Omega_{\rm o}$ parameter space, as well as explore the role that different

229: axial boundary conditions, on both the flow \citep{KJG04} and the field

230: \citep{LIU06}, might play.  Future numerical work will similarly consider

231: the problem in bounded cylinders, with different boundary conditions on the

232: endplates.  Only then will we fully understand what influence the endplates,

233: and possible asymmetries between them, have on the MRI.

234: 

235: \acknowledgments

236: 

237: This work was supported by the German Leibniz Gemeinschaft, under program

238: SAW.

239: 

240: \begin{thebibliography}{}

241: 

242: \bibitem[Balbus \& Hawley(1991)]{BH91} Balbus, S. A. \& Hawley, J. F.

243: 1991, \apj, 376, 214

244: \bibitem[Hollerbach \& Fournier(2004)]{HF04} Hollerbach, R. \& Fournier, A.

245: 2004, In Rosner et al., p.\ 114 (see also astro-ph/0506081)

246: \bibitem[Hollerbach \& R\"udiger(2005)]{HR05} Hollerbach, R. \& R\"udiger,

247: G. 2005, Phys.\ Rev.\ Lett., 95, 124501

248: \bibitem[Ji et al.(2001)]{JGK01} Ji, H. T., Goodman, J. \& Kageyama, A.

249: 2001, \mnras, 325, L1

250: \bibitem[Kageyama et al.(2004)]{KJG04} Kageyama, A., Ji. H. T., Goodman, J.,

251: Chen, F. \& Shoshan, E. 2004, J. Phys.\ Soc. Japan, 73, 2424

252: \bibitem[see also Knobloch(1996)]{Knobloch} Knobloch, E., 1996,

253: Phys.\ Fluids, 8, 1446

254: \bibitem[Liu et al.(2006)]{LIU06} Liu, W., Goodman, J., Herron, I. \&

255: Ji, H. T. 2006, Phys.\ Rev.\ E, submitted

256: \bibitem[Rosner et al.(2004)]{RRB04} Rosner, R., R\"udiger, G. \&

257: Bonanno, A. (Editors) 2004, MHD Couette Flows: Experiments and Models,

258: AIP Conf.\ Proc.\ 733

259: \bibitem[R\"udiger \& Zhang(2001)]{RZ01} R\"udiger, G. \& Zhang, Y.  2001,

260: A\&A, 378, 302

261: \bibitem[R\"udiger et al.(2005)]{RHS05} R\"udiger, G., Hollerbach, R.,

262: Schultz, M. \& Shalybkov, D. A.  2005, Astron.\ Nachr., 326, 409

263: \bibitem[Stefani et al.(2006)]{SGGRSSH} Stefani, F., Gundrum, T., Gerbeth,

264: G., R\"udiger, G., Schultz, M., Szklarski, J. \& Hollerbach, R.  2006,

265: Phys.\ Rev.\ Lett., submitted (see also astro-ph/0606473)

266: \bibitem[Tobias et al.(1998)]{TPK98} Tobias, S. M., Proctor, M. R. E.,

267: Knobloch, E. 1998, Physica D, 113, 43

268: 

269: \end{thebibliography}

270: \end{document}

271: 

272: