1: \documentstyle[pre,aps,multicol,epsf]{revtex}  

2: %\input REVTeXDraft1.tex 

3: \def\B.#1{{\bbox{#1}}}  

4: \def\C.#1{{\cal{#1}}} 

5: \def\pop#1{{{\partial}\over{\partial#1}}} 

6: \def\popa#1#2{{{\partial#1}\over{\partial#2}}} 

7: \def\p#1#2{\partial#1/\partial#2} 

8: \def\i{{\rm i}}

9: \def\d{{\rm d}} 

10: \def\D{{\rm D}} 

11: \def\invr{O\left(R^{-1}\right)} 

12: \def\siminvr{\sim O\left(R^{-1}\right)} 

13: \def\abs#1{\left|#1\right|} 

14: \def\ep{{\epsilon}} 

15: \def\be{\begin{equation}} 

16: \def\ee{\end{equation}} 

17: \def\etal{{\em et al.} } 

18: \def\alpha{k} 

19: \def\Re{{\rm Re}} 

20: \def\Rec {${\rm Re}_{\rm cr}$}

21: \begin{document} 

22: \title{{\rm PRL} \hfill {\sl Submitted}\\ Retardation of the Onset of

23: Turbulence by Minor Viscosity Contrasts} \author{Rama

24: Govindarajan$^{\dag,*}$, Victor S. L'vov$^*$ and Itamar Procaccia$^*$}

25: \address{$^*$ Dept. of Chemical Physics, The Weizmann Institute of

26: Science, Rehovot 76100, Israel \\ $^\dag $ Fluid Dynamics Unit,

27: Jawaharlal Nehru Centre for Advanced Scientific Research, Jakkur,

28: Bangalore 560064, India.} \maketitle

29: %%%%%%%%%%%%%%%%%%%%%% 

30:  \begin{abstract}  

31: Motivated by the large effect of turbulent drag reduction by minute

32: concentrations of polymers we study the effects of minor viscosity

33: contrasts on the stability of hydrodynamic flows. The key player is a

34: localized region where the energy of fluctuations is produced by

35: interactions with the mean flow (the ``critical layer"). We show that

36: a layer of weakly space-dependent viscosity placed near the critical

37: layer can have very large stabilizing effect on hydrodynamic

38: fluctuations, retarding significantly the onset of turbulence. The

39: effect is {\em not} due to a modified dissipation (as is assumed in

40: theories of drag reduction), but due to reduced energy intake from the

41: mean flow to the fluctuations.  We propose that similar physics act in

42: turbulent drag reduction.

43: \end{abstract}

44:  

45: \begin{multicols}{2} 

46:  

47: The addition of small amounts of polymers to hydrodynamic systems

48: produces dramatic effects on phenomena such as the transition to

49: turbulence, vortex formation and turbulent transport \cite{95NH}. The

50: phenomenon that attracted most attention was, for obvious reasons, the

51: reduction of friction drag by up to 80\% when a very small

52: concentration of long-chain polymers were added to turbulent flows

53: \cite{69Lum,00SW}. In spite of the fact that the phenomenon is robust

54: and the effect huge, there exists no accepted theory that can claim

55: quantitative agreement with the experimental facts.  Moreover, it

56: appears that there is no mechanistic explanation.  In the current

57: theory that is due to de Gennes \cite{86TG,90Gennes} one expects the

58: Kolmogorov cascade to be terminated at scales larger than Kolmogorov

59: scale, leading somehow to an increased buffer layer thickness and

60: reduced drag, but how this happens and what is the fate of the

61: turbulent energy is not being made clear.

62: 

63: In this Letter we propose that the crucial issue is in the {\em

64: production} of energy of hydrodynamic fluctuations by their

65: interaction with the mean flow.  For the sake of concreteness we

66: examine a simple laminar flow and its loss of stability, and show how

67: small viscosity contrasts lead to an order of magnitude retardation in

68: the onset of instability of ``dangerous" disturbances. In this model

69: everything is explicitly calculable, and we demonstrate that nothing

70: special happens to the dissipation. Rather, it is the energy

71: production that is dramatically reduced, giving rise to a large effect

72: for a small cause. At the end of this Letter we argue that similar

73: physics may very well be at the heart of turbulent drag reduction, but

74: we stress that the phenomenon discussed below is interesting by itself

75: and well warrants an experimental confirmation.

76: 

77: It is well known that parallel Poiseuille flow loses its stability at

78: some threshold Reynolds number Re=Re$_{\rm th}$ (close to 5772).  It

79: is also well known that the instability is ``convective", with the

80: most unstable mode having a phase velocity $c$. Analytically it has

81: the form

82: \begin{equation} 

83:   \label{eq:inst}  

84:    \hat \phi(x,y,t) = \case{1}{2}\{\phi(y) 

85: \exp [\i \alpha(x-c \, t)] + 

86:    \mbox{c.c.} \} \exp(\gamma t) \ , 

87: \end{equation}

88: where $\hat \phi(x,y,t)$ is the disturbance streamfunction and

89: $\phi(y)$ is the complex envelope of $\hat \phi(x,y,t)$.  We have

90: chosen

91: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

92: \begin{figure} 

93: \epsfxsize=7.5cm \epsfbox{Ramaj1.eps}

94: \vskip 0.3cm  

95: \caption{Schematic of the flow: the fluid near the walls has a 

96: viscosity $\mu_1$, and that flowing at the center is of

97: viscosity $\mu_2$. In the mixed layer

98: (of width $q$) the viscosity varies gradually between $\mu_1$ and $\mu_2$. The

99: parameter $p$ controls the position  of the mixed layer.}

100:  \label{f:scheme} 

101: \end{figure} 

102: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   

103: \noindent

104:  $x$ and $\alpha$ as the streamwise coordinate and wavenumber of

105: the disturbance, $c$ as the phase speed and $t$ as time. $\gamma$ is

106: the increment of instability.  What is not usually emphasized is that

107: the main interactions leading to the loss of stability occur in a

108: sharply defined region in the channel, i.e. at a layer whose distance

109: from the wall is such that the phase velocity $c$ is comparable to the

110: velocity of the mean flow. We refer to this layer as the ``critical"

111: layer.  It is thus worthwhile to examine the effect on the stability

112: of Poiseuille flow of a viscosity gradient placed in the vicinity of

113: the critical layer. Following \cite{rama} we examine a channel flow of

114: two fluids with different viscosities $\mu_1$ and $\mu_2$, see

115: Fig. \ref{f:scheme}.

116: 

117: Observing that the inferred effective viscosity in polymer drag

118: reduction increases towards the center by about 30\% over about a 1/3

119: of the half-channel \cite{97SBH}, we choose $\mu_2=1$ and $m\equiv

120: \mu_1/ \mu_2=0.9$, with all the viscosity difference of 0.1

121: concentrated in a ``mixing" layer of width 0.1, leading to comparable

122: viscosity gradients. The observation that we want to focus on is shown

123: in Fig. 2: the threshold Reynolds number for the loss of stability of

124: the mode as in Eq. (\ref{eq:inst}) depends crucially on the position

125: of the mixing layer. When the latter hits the critical layer the

126: threshold Reynolds number for the loss of stability reaches as much as

127: 88000. In other words, one can increase the threshold of instability

128: {\em for a given mode} 15 times, and by making the mixing layer

129: thinner one can reach even higher threshold Reynolds values.

130: %%%%%%%%%%%%%%%%%%%%%%%

131: \begin{figure}

132: \epsfxsize=8.3 cm

133: \epsfbox{Ramarcr.eps}

134: \caption{The dependence of the threshold Reynolds number on the

135: position of the viscosity stratified layer for $m=0.9$. The dashed

136: line pertains to the neat fluid. Note the huge increase in $R_{\rm

137: th}$ within a small range. This occurs when the stratified layer

138: overlaps the critical layer.}

139: \end{figure}

140: %%%%%%%%%%%%%%%%%%%%%%%

141: In this Letter we analyze the physical origin of this huge sensitivity

142: of the flow stability to the profile of the viscosity.  The stability

143: of this flow is governed by the modified Orr-Sommerfeld equation

144: \cite{white}

145: \begin{eqnarray}\nonumber  

146: &&  \i\alpha \left[\left(\phi''-\alpha^2\phi\right)(U-c-\i \gamma)  

147: - U''\phi \right] = 

148: {1\over \Re}\bigg[\mu \phi^{\rm (4)}+ 2 \mu' \phi''' \\ \label{modOS} 

149: && +\left(\mu'' - 2 \alpha^2 \mu \right)\phi'' - 2 \alpha^2 \mu' 

150: \phi' + \left(\alpha^2 \mu'' + \alpha^4 \mu\right) \phi\bigg]\,,  

151: \end{eqnarray} 

152: in which $\phi$, $U$ and $\mu$ are functions of $y$. The boundary

153: conditions are $\phi(\pm 1) = \phi' (\pm 1) = 0$.  All quantities have

154: been non-dimensionalised using the half-width $H$ of the channel and

155: the centerline velocity $U_0$ as the length and velocity scales

156: respectively.  The Reynolds number is defined as $\Re \equiv \rho U_0

157: H /\mu_2$, where $\rho$ is the density (equal for the two fluids).

158: The primes stand for derivative with respect to $y$.  At $y=0$, we use

159: the even symmetry conditions $\phi(0)=1$ and $\phi'(0)=0$, as the even

160: mode is always more unstable than the odd.

161:  

162: 

163: 

164: Since the flow is symmetric with respect to the channel centerline, we

165: restrict our attention to the upper half-channel. Fluid 2 occupies the

166: region $0 \le y \le p$. Fluid 1 lies between $p+q \le y \le 1$. The

167: region $p \le y \le p+q$ contains mixed fluid. The viscosity is

168: described by a steady function of $y$, scaled by the inner fluid

169: viscosity $\mu_2$:

170: \begin{eqnarray}\label{muin}

171:    \mu(y) &=& 1 \,,\quad  \mbox{for} \quad 0 \le y \le p\,,\\

172: \label{vis5}

173: \mu(y) &=& 1 + (m-1)\, \xi^3\left[10 - 15\, \xi + 6 \xi^2

174: \right], \ 0 \le \xi \le 1\,, \\

175: \label{muout}

176:    \mu(y) &=& m \,,\quad  \mbox{for} \quad p+q \le y \le 1\,,\\

177: \end{eqnarray}

178: Here $\xi\equiv (y-p)/q$ is the mixed layer coordinate.  We have

179: assumed a 5th-order polynomial profile for the viscosity in the mixed

180: layer, whose coefficients maintain the 

181: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

182: \begin{figure} 

183: \epsfxsize=7.5cm  \epsfbox{RamaU.eps} 

184: \caption{Profiles of the normalized viscosity $\mu(y)$ 

185: and normalized velocity $U(y)$ and the second derivative $U^{''}(y)$

186: for $m=0.9$ (solid lines) and $m=1.0$ (dashed lines).  The mixed layer

187: is between the vertical dashed lines. }

188: \label{f:prof} 

189: \end{figure} 

190: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

191: \noindent

192: viscosity and its first two derivatives continuous across the mixed

193: layer.  The exact form of the profile is unimportant. For a plot of

194: the profile $m=0.9$, see Fig.~\ref{f:prof}.

195: 

196: The basic flow $U(y)$ is obtained by requiring the velocity and all

197: relevant derivatives to be continuous at the edges of the mixed layer:

198: \begin{eqnarray}\label{ui}

199: &   U(y) = 1-Gy^2/2\,, \quad &\mbox{for} \quad y \le p\,,\\

200: &U(y) = U(p) - G\int_p^y{y\over \mu} \d y\,,\quad   &\mbox{for}

201: \quad p \le y \le p+ q,

202: \label{um}\\

203: & U(y) = G\left(1-y^2\right)/2m, \quad  &\mbox{for}

204: \quad y \ge p+q\ .\label{uo}

205: \end{eqnarray}

206: Here $G$ is the streamwise pressure gradient.  

207: 

208: It can be seen, comparing the mean profile $U(y)$ to that of the neat

209: fluid (cf. Fig. \ref{f:prof}), that nothing dramatic happens to this

210: profile even when the mixing layer is chosen to overlap a typical

211: critical layer. Accordingly we need to look for the origin of the

212: large effect of Fig. 2 in the energetics of the disturbances. To do

213: so, recall that the streamwise and normal components of the

214: disturbance velocity $\hat u(x,y,t)$ and $\hat v(x,y,t)$ may be

215: expressed via streamfunction as usual:

216: \begin{equation} 

217:   \label{velcomp} 

218: \hat u(x,y,t) = \partial\hat \phi/\partial y \quad {\rm and} \quad \hat 

219: v(x,y,t) = -\partial \hat \phi/\partial x \ . 

220: \end{equation} 

221: These functions may be written in terms of complex envelopes similar 

222: to Eq.~(\ref{eq:inst}): 

223: \begin{eqnarray} 

224:   \label{eq:vel-env} 

225:   \hat u(x,y,t) &=& \case{1}{2}\big\{u(y) \exp\left[\i\alpha(x-c\,

226:   t)\right] + \mbox{c.c.}\big\} \exp(\gamma t) \,,\\ \nonumber \hat

227:   v(x,y,t) &=& \case{1}{2}\big\{v(y) \exp\left[\i\alpha(x-c\,

228:   t)\right] + \mbox{c.c.}\big\} \exp(\gamma t) \ .

229: \end{eqnarray} 

230: The pressure disturbance $\hat p$ is defined similarly. 

231: 

232: Define now a disturbance of the density of the kinetic energy 

233: \begin{equation}\label{kinen} 

234:   \hat E(x,y,t) = \case{1}{2}\left[ \hat u(x,y,t)^2 + \hat 

235:   v(x,y,t)^2\right] 

236: \end{equation} 

237: we can express the mean (over $x$) density of the kinetic energy as

238: follows:

239: \begin{eqnarray}\label{averen} 

240:   E(y,t)&\equiv& \left< \hat E(x,y,t)\right>_x =\C.E (y)\exp\,(2 

241: \gamma t)\,,

242: \\ \nonumber

243: \C.E (y)&=&\case{1}{4}\left( |u(y)|^2 + |v(y)|^2 \right)\ . 

244:   \end{eqnarray}

245: 

246: The physics of our phenomenon will be discussed in terms of the

247: balance equation for the averaged disturbance kinetic energy. Starting

248: from the linearized Navier-Stokes equations for $\hat u$ and $\hat v$,

249: dotting it with the disturbance velocity vector, averaging over one

250: cycle in $x$ and using Eqs. (\ref{eq:vel-env})-(\ref{averen}) leads to

251: \be 2 \gamma \,\C.E (y) = \nabla \cdot J(y) + W_+(y) - W_-(y) \,,

252: \label{enbal} 

253: \ee where the energy flux $J(y)$ in the $y$ direction, rates of energy

254: production (by the mean flow) $W_+(y)$ and energy dissipation (by the

255: viscosity) $W_-(y)$ are given by

256: \begin{eqnarray}\label{current} 

257:   J(y) &\equiv&  {\left[u(y) p^*(y)+\mbox{c.c.}\right] 

258: \over 4 \rho} + {1 \over \Re} \mu (y) \nabla \C.E(y)\,,\\

259: W_+(y) &\equiv&  - {1 \over 4} U'(y)\left[u(y)

260:  v^*(y) +\mbox{c.c.}  \right]\,,

261: \label{prod} \\

262: W_-(y) &\equiv& {\mu(y) \over \Re} \left\{2 \alpha^2 \C.E(y)

263: + {1 \over 2}\left[|u'(y)|^2+|v'(y)|^2\right]\right\}\ .

264: \label{diss}

265: \end{eqnarray}

266: The superscript $*$ denotes the complex conjugate. To plot these

267: functions we need to solve Eq. (\ref{modOS}) as an eigenvalue problem,

268: to obtain $c$, $\gamma$, and $\phi(y)$ at given Re and $k$. The value

269: of $c$ determines the position of the critical layer.

270: 

271: It is convenient to compute and compare the space averaged production

272: and dissipation terms $\Gamma_+$ and $\Gamma_-$ defined by:

273: %%%%%%%%%%%%%%%%%%

274: \begin{equation}\label{totpd}

275:   \Gamma_\pm

276:  \equiv\int_0^1 W_\pm(y) \d y \ \Big/ \int_0^1 \C.E (y) \d y

277:  \ .

278: \end{equation}

279: The local production of energy can be positive or negative, indicative

280: of energy transfer from the mean flow to the disturbance and vice-versa

281: respectively. The production in one region (where $ W_+(y)>0$) can be

282: partly canceled out by a ``counter-production'' in other region (where

283: $ W_+(y)<0$). 

284: 

285: The use of these measures can be exemplified with the neat fluid

286: ($m=1.0$ here). The laminar flow displays its first linear instability at

287: a threshold Reynolds number of $Re_{\rm th}=5772$, which means that

288: the total production $\Gamma_+$ across the layer becomes equal to the 

289: total dissipation $\Gamma_-$ at this value of $\Re$. Examining Fig. 

290: \ref{f:far} we can see that the disturbance kinetic energy is produced 

291: predominantly within the 

292: critical layer, where the basic flow velocity is close to the phase 

293: speed of the disturbance, while most of the dissipation is in the wall 

294: layer. 

295: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

296: \begin{figure} 

297: \epsfxsize=7.5cm \epsfbox{Ramas0.eps}  

298: \vskip -1cm

299: \epsfxsize=7.5cm \epsfbox{Ramafar.eps}  

300: \caption{Energy balance: production $W_+(y)$, solid line; 

301: dissipation $W_-(y)$, dot-dashed line, $\Re=5772$. Top: $m=1$,

302: $\Gamma_+=\Gamma_-=0.0148$. Bottom: $m=0.9$, $p=0.3$,

303: $\Gamma_+=0.0158$, $\Gamma_-=0.0148$. In this and all the subsequent

304: figures the solid vertical lines show the location of the critical

305: lines, whereas the region between the dotted lines is the mixed

306: layer.}

307:  \label{f:far} 

308: \end{figure} 

309: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%  

310: The balance is not changed significantly when the viscosity ratio is

311: changed to $0.9$ so long as the mixed layer is not close to the

312: critical layer. There is a small region of production and one of

313: counter-production within the mixed layer, whose effects cancel out,

314: leaving the system close to marginal stability.

315: 

316: We now turn our attention to Fig.~\ref{f:near}, in which the main

317: point of this Letter is demonstrated.  Here, the Reynolds number is

318: the same as before, but the mixing layer has been moved close to the

319: critical layer. It is immediately obvious that the earlier balance is

320: destroyed. The counter-production peak in the mixed layer is much

321: larger than before, making the flow more stable.  The wavenumber used

322: is that at which the flow is least stable for the given Reynolds

323: number at this $p$.  For $m=0.9$, the threshold Reynolds number is

324: $46400$.  Fig. \ref{f:marg} shows the energy balances at marginal

325: stability - the picture is qualitatively the same here as at

326: $Re\approx 5772$ for the neat fluid.

327: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

328: \begin{figure} 

329: \epsfxsize=7.5cm \epsfbox{Ramanear.eps}

330: 

331: \epsfxsize=7.5cm \epsfbox{Ramamarg.eps} 

332: \caption{Energy balance for $m=0.9$, $p=0.85$. Energy 

333: roduction $W_+(y)$, solid line; dissipation $W_-(y)$, dot-dashed line.

334: Upper panel: stable flow at $\Re=5772$ 

335: (with $\Gamma_+= -0.0114$, $\Gamma_-=0.0122$).

336: Lower panel: Marginal flow at  $\Re=46400$ 

337: (with $\Gamma_+=\Gamma_-=0.0053$). }

338:  \label{f:near} 

339: \end{figure} 

340: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%   

341: The main factor determining the stability is the production, which is

342: driven by the phase change caused by the viscosity stratification.

343: The dissipation on the other hand depends only on Reynolds number and

344: does not respond disproportionately to changes in viscosity.  In neat

345: fluids, the term containing $U''(y)$ in (\ref{modOS}) is always of

346: higher order within the critical layer. However, with the introduction

347: of a viscosity gradient within the critical layer, the gradients of

348: the basic velocity profile will scale according to the mixed layer

349: coordinate $\xi$.  An analysis in the critical layer indicates that

350: for $q \le O(Re^{-1/3})$, the term containing $U''/(U-c)$ is now among

351: the most dominant. Any reasonable viscosity gradient of the right sign

352: will pick up this term, leading to vastly enhanced stability. Note the

353: dramatic effect in $U''$ in Fig. 3.

354: 

355: Indeed, in the light of this discussion we can expect that the large

356: effect of retardation of the instability would even increase if we

357: make the mixing layer thinner. This is indeed so. Nevertheless, one

358: cannot conclude that instability can be retarded at will, since other

359: disturbances, differing from the primary mode, become unstable first,

360: albeit at a much higher Reynolds number than the primary mode; when we

361: stabilize a given mode substantially, we should watch out for other

362: pre-existing/newly destabilized modes which may now be the least

363: stable.

364: 

365: Finally, we connect our findings to the phenomenon of drag reduction

366: in turbulent flows. Since the total dissipation can be computed just

367: from the knowledge of the velocity profile at the walls, any amount of

368: drag reduction must be reflected by a corresponding reduction of the

369: gradient at the walls.  Concurrently, the energy intake by the

370: fluctuations from the mean flow should reduce as well. Indeed, the

371: latter effect was measured in both experiments \cite{97THKN} and

372: simulations \cite{01DSBH,00Ang}.  The question is which is the chicken

373: and which is the egg. In our calculation we identified that the

374: reduction in production comes first.  From Figs. 4 and 5 (upper panel)

375: which are at the same value of Re we see that the dissipation does not

376: change at all when the mixing layer moves, but the production is

377: strongly affected. Of course, at steady state the velocity gradient at

378: the wall must adjust as shown in Fig.~5 (lower panel). We recognize

379: that in a turbulent flow there are a number of modes that interact,

380: but we propose that a similar mechanism operates for each mode at its

381: critical layer, where both elastic and viscous effect determine the

382: mean flow. In the present calculation we can consider all the putative

383: unstable modes, and conclude that with a viscosity gradient similar to

384: that seen in polymeric turbulent flows the threshold Re goes up five

385: times (to 31000).  We leave the confirmation of this prediction to

386: future experiments.

387: 

388:   {\bf Acknowledgments}. This work has been supported by the Israel

389:  Science Foundation.

390: \begin{references} 

391: 

392: \bibitem{95NH} R.H. Nadolink and W.W. Haigh, ASME

393: Appl. Mech. Rev. {\bf 48}, 351 (1995).

394: 

395: \bibitem{69Lum} J. L. Lumley, Ann. Rev. Fluid Mech. {\bf 1}, 367

396: (1969).

397: 

398: \bibitem{00SW} K.R. Sreenivasan and C. M. White, J. Fluid Mech., {\bf

399: 409}, 149 (2000).

400: 

401: \bibitem{86TG} M. Tabor and P.G. de Gennes, Europhys. Lett., {\bf 2},

402: 519 (1986).

403: 

404: \bibitem{90Gennes} P.G. de Gennes, {\em Introduction to Polymer

405: Dynamics}, (Cambridge, 1990).

406: 

407: \bibitem{rama} B. T. Ranganathan and R. Govindarajan, Phys. Fluids

408: {\bf 13}, 1 (2001).

409: 

410: \bibitem{97SBH} R. Sureshkumar, A. N.  Beris and R.A. Handler,

411: Phys. Fluids {\bf 9}, 743 (1997).

412: 

413: \bibitem{white} F.M White, {\em Viscous Fluid Flow} (McGraw Hill,

414: 1991).

415: 

416: \bibitem{97THKN} J.M.J den Toonder, M.A. Hulsen,G.D.C Kuiken and F.T.M

417: Niewstadt, J. Fluid Mech. {\bf 337}, 193 (1997).

418: 

419: \bibitem{01DSBH} C.D. Dimitropoulos, R. Sureshkumar, A.N. Beris and

420: R.A. Handler, Phys. Fluids, {\bf 13}, 1016 (2001).

421: 

422: \bibitem{00Ang} E. De Angelis, ``The Influence of Polymer Additives on

423: the Structure of Wall Turbulence", Thesis, Universita' di Roma, La

424: Sapienza, 2000.

425: 

426: \end{references}  

427:  

428: \end{multicols} 

429: %%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% 

430: \end{document}

431: 

432: 

433: