1: \documentclass[12pt]{article}

2: \usepackage[cp866]{inputenc}

3: \usepackage[russian]{babel}

4: \usepackage{graphicx}

5: %\documentstyle[12pt,cyrwin%

6: %Russian

7: %]{article}

8: %\documentclass [12pt]{article}

9: \begin{document}

10: \addtolength{\topmargin}{-\topmargin}

11: \addtolength{\topmargin}{.1in}\addtolength{\textheight}{.2\textheight}

12: \addtolength{\headheight}{-\headheight}\addtolength{\headsep}{-\headsep}

13: \oddsidemargin 0.cm\hsize 160mm

14: \raggedbottom\renewcommand{\baselinestretch}{1.2}

15: \newcommand{\rot}{\mathop{\rm rot}\nolimits}

16: \newcommand{\grad}{\mathop{\rm grad}\nolimits}

17: \newcommand{\diver}{\mathop{\rm div}\nolimits}

18: \newcommand{\km}{\mathop{\rm km}\nolimits}

19: \newcommand{\s}{\mathop{\rm s}\nolimits}

20: \newcommand{\keV}{\mathop{\rm keV}\nolimits}

21: \newcommand{\eV}{\mathop{\rm eV}\nolimits}

22: \newcommand{\Ma}{\mathop{\rm Ma}\nolimits}

23: \newcommand{\Fr}{\mathop{\rm Fr}\nolimits}

24: \newcommand{\const}{\mathop{\rm const}\nolimits}

25: \newcommand{\Erf}{\mathop{\rm Erf}\nolimits}

26: \newcommand{\Ree}{\mathop{\rm Re}\nolimits}

27: \normalsize

28: 

29: \centerline{\bf EXACT EXPANSION LAW}

30: 

31: \centerline{\bf FOR RICHTMYER-MESHKOV TURBULENT MIXING ZONE}

32: 

33: N.A. Inogamov$^1$ \& A.M. Oparin$^2$

34: 

35: 1. Landau Institute for Theoretical Physics, Moscow

36: 

37: 2. Institute for Computer-Aided Design, Moscow

38: 

39: \centerline{Abstract}

40: 

41: {\it Definition of mixing fronts and turbulent mixing zone.}

42: Development of the Richtmyer-Meshkov instability

43: leads to mixing of two substances

44: separated by contact boundary.

45: Let the axis $z$ is perpendicular to a plane of boundary

46: and $z=0$ corresponds to an unperturbed plane.

47: The turbulent mixing zone is limited by front $z=h_+(t)>0$

48: from heavy fluid side

49: and by front $z=h_-(t)<0$ from light fluid.

50: Mixture of two fluids is confined between these two fronts

51: (layer $h_-<z<h_+).$

52: 

53: 

54: {\it Initial conditions.}

55: Before pass of a shock wave both liquids were still

56: [were at rest: $\vec v(x,y,z,t<0)\equiv 0$]

57: and $z=\eta_{ini}(x,y)$ was perturbed contact boundary.

58: We consider linear (small amplitude) perturbation: $|\nabla\eta_{ini}|\ll 1.$

59: In this case the shock wave passes perturbations quickly:

60: $t_s\ll \lambda_{ini}/|\dot\eta_{\lambda ini}|,$

61: where $t_s$ is shock wave passage (sonic) time,

62: $\lambda_{ini}$ is a typical wavelength (or space scale) of perturbations,

63: and $|\dot\eta_{\lambda ini}|$ is typical velocity after shock wave passage.

64: 

65: 

66: {\it Transition and asymptotic stages.}

67: Universal asymptotics

68: follows the transition stage at late time

69: $t\gg \lambda/|\dot\eta_{\lambda ini}|.$

70: Functions $h_\pm(t)$ asymptoticaly transform to power law dependences.

71: Mixing is accompanied by cascade of enlarging of dominant scale

72: of a turbulent mixing zone.

73: Under condition $\mu=0$ exact scaling laws are:

74: $h_+^{2D}\propto t^{2/5}$ and $h_+^{3D}\propto t^{1/3},$

75: where $\mu$ is density ratio,

76: 2D and 3D mark dimension of space.

77: These scalings are calculated in our work.

78: The scaling exponents 2/5 and 1/3

79: are determined by the mechanism of redistribution

80: of $z-$component of momentum from small into the large scales.

81: 

82: For $\mu\neq 0$ the exponents 2/5 and 1/3

83: define the bottom limit of a range of possible values

84: for both functions $h_+(t)$ and $h_-(t).$

85: Our 2D direct numerical simulations at different values $\mu\sim 1$

86: show that $d\ln h_\pm/dt\approx 2/5.$

87: 

88: {\bf This work has been supported by RBRF (grants 02-02-17499, 03-01-00700)

89: and scientific schools (NSh-2045.2003.2, NSh-70.2003.1).}

90: 

91: 

92: 

93: \newpage

94: \centerline{{\bf \# 1. Introduction}}

95: 

96: \vspace{.5cm}

97: The Richtmyer-Meshkov instability

98: starts to develop after crossing (passage) by a shock wave of a contact boundary.

99: In case of small on amplitude initial perturbation

100: evolution in time proceeds in three stages

101: (we shall designate them I, II and III).

102: At short acoustic stage I\footnote{

103: In case of small initial perturbations

104: "hierarchy" of characteristic times for the stages I, II and III

105: is following:

106: $t_I\ll t_{II}\ll t_{III}$.

107: }

108: formation of a near boundary velocity field completes.

109: This field appears due to interaction of contact boundary with a shock wave.

110: Near boundary field defines the further hydrodynamical\footnote{

111: In case of small initial perturbations

112: hydrodynamical motion of substance at stages II and III

113: occur to velocities small in comparison with sonic velocity

114: (incompressible motion).

115: At stage I at presence of a shock wave

116: with Mach number Ma significantly larger than 1

117: compressibility effects are essential.

118: } motion.

119: 

120:  

121: Duration of an acoustic stage is about

122: $$

123: t_s \sim \lambda_{ini}/c_s^{min}. \eqno (1)

124: $$

125: Where $\lambda_{ini}$ is typical transverse scale of initial perturbations,

126: $c_s^{min}$ is minimal from sonic velocities after shocks

127: or after expansion wave.

128: In case of usual equation of state this is sonic velocity in heavy gas

129: because pressure is continuous at contact boundary

130: and $c_s\sim \sqrt{p/\rho}$.

131: 

132: 

133: The expression (1) includes space scale $\lambda_{ini}$

134: (not amplitude $\eta_{\lambda ini}$)

135: as acoustic interaction of the curved contact boundary

136: with the curved shock waves continues

137: while shock waves will not depart from boundary on the large distances

138: $>\lambda_{ini}$.

139: Interaction is carried out by means of a series of oblique sound waves

140: which move in the expanding layer between shocks and boundary.

141: This interaction

142: creates vorticity which defines motion at stages II and III\footnote{

143: Baroclinic generation of vorticity at the boundary

144: and generation of vorticity in the bulk by curved shocks

145: (Schilling et al., 2004; Wouchuk, 2001a, 2001b).

146: In the case of weak shocks we can neglect bulk vorticity.

147: }.

148: 

149: 

150: At a stage II near boundary velocity field decays

151: at removal from boundary on distances which exceed $\lambda_{ini}$.

152: The law of reduction in velocity in depth from boundary

153: depends on transversal correlations of perturbations.

154: This law qualitatively differs

155: for periodic, solitary and random perturbations\footnote{

156: Exponential decay for periodic perturbations

157: and power laws for random and solitary perturbations

158: at the same effective depth $\sim\lambda_{ini}$ of near boundary velocity field.}.

159: In turn decay of velocity in depth

160: determines the law of expansion of a zone of turbulent mixing at a stage III.

161: The given work

162: is devoted to the analysis of distribution of velocity near boundary.

163: 

164: 

165: 

166: At stages II and III

167: entropy and vortical modes on the one hand

168: and acoustic modes with another - are decoupled\footnote{

169: They are very different in velocities.

170: Entropy and vortical modes are slow

171: while acoustic modes are fast.

172: }.

173: Therefore an influence of compressibility becomes nonsignificant.

174: Duration of sonic stage (1) is small in comparison with main time scale

175: $$

176: t_{\lambda ini}\sim \lambda_{ini} /w_{\lambda ini}. \eqno (2)

177: $$

178: The time (2) is necessary

179: for development of perturbations up to nonlinear or "saturated" amplitudes.

180: Displacement of boundary during short time (1) is small

181: in comparison with $\lambda_{ini}$.

182: 

183: 

184: An estimate based on the Richtmyer formula gives

185: $$

186: w_{\lambda ini}\sim

187:  {\rm At}^+ k_{ini} \eta_{\lambda ini}^+ W_{unpert CB}, \eqno (3)

188: $$

189: where At$^+$ is Atwood number after (sign plus) shock,

190: $k_{ini}=2\pi/\lambda_{ini},$

191: $\eta_{\lambda ini}^+$ is typical amplitudes of boundary perturbations

192: (after shock)

193: when we shift at distance $\sim \lambda_{ini}$ in transverse direction,

194: $W_{unpert CB}$ is unperturbed velocity of contact after shock,

195: $W_{unpert CB}\sim c_s.$

196: From equation (3) follows,

197: that a smallness of $t_s$ (1)

198: in comparison with $t_{\lambda ini}$ (2)

199: it is caused by linearity of initial perturbations.

200: 

201: 

202: 

203: Let

204: $$

205: \eta_{\lambda ini}^- \sim \lambda_{ini} \eqno (4)

206: $$

207: (nonlinear initial perturbations),

208: here $\eta_{\lambda ini}^-$ is amplitude of surface perturbation before a shock.

209: In this case we cannot separate stages I and II:

210: $t_s \sim t_{\lambda ini},$

211: $w_{\lambda ini}\sim W_{unpert CB}\sim c_s.$

212: Hence, the stage III "will start"

213: with high initial velocities

214: making an appreciable share from sonic velocity.

215: Nevertheless,

216: received below asymptotical expansion laws

217: of a turbulent mixing zone

218: do not vary

219: in case of large perturbation (4)

220: in comparison with a case of linear perturbations.

221: The matter is that

222: eventually velocities are reduced with time,

223: and become small in comparison with $c_s$.

224: 

225: 

226: 

227: 

228: 

229: 

230: 

231: Motion at a stage II is caused by movement of the primary vorticies

232: $\sim\lambda_{ini}$

233: created by a shock from surface perturbations at a sonic stage.

234: As we will see,

235: the essence of motion comes to light

236: at introduction of concept about generations

237: and terms primary, secondary, tertiary, etc.

238: The question is the current generation of a chain

239: of sign-variable vorticies

240: of the size $\lambda_{curr}(t)$

241: and about evolution of this generation.

242: Such chain refers to also as dominating or large-scale structure.

243: Evolution consists in growth of the size $\lambda_{curr}(t)$

244: as a result of merging of vorticies.

245: Each stage of merging

246: connects parental generation

247: and generation of first "descendants" of these "parents".

248: Asymptotical law of expansion

249: is established for far "descendants" of primary vorticies.

250: Motion at a stage II has transitive character and is not universal.

251: 

252: 

253: 

254: 

255: 

256: 

257: Fundamental interest represents

258: a finding of the law of expansion

259: of a zone of mixing at large times $t\gg t_{\lambda ini}$.

260: This is an asymptotical turbulent stage

261: (stage III)

262: of development of the Richtmyer-Meshkov instability.

263: Appears, present universal asymptotics, describing a stage III.

264: Its representation makes the basic contents of the given work.

265: As against laws of mixing

266: at Rayleigh-Taylor and Kelvin-Helmholts turbulence

267: when exists simple self-similarity,

268: here simple self-similarity is not present.

269: In Rayleigh-Taylor and Kelvin-Helmholts cases

270: there is a unique dimensional parameter:

271: free fall acceleration $g$

272: or a velocity difference $\Delta u$ accordingly.

273: In the Richtmyer-Meshkov case

274: there are two independent parameters

275: $\lambda_{ini}$ and $w_{\lambda ini}$.

276: Therefore simple self-similarity is absent.

277: At such position presence universal asymptotics

278: is rather not trivial bright phenomenon.

279: The mechanism is below described,

280: responsible for expansion of Richtmyer-Meshkov turbulent zone

281: with self-similar hierarchy of generations.

282: 

283: 

284: 

285: 

286: At stage III

287: during each moment of time

288: in a mixing layer

289: the vortical structure dominates

290: over the current scale $\lambda_{curr}(t) \sim h(t)$,

291: which grows with $t$

292: $(h$ is thickness of a layer of interpenetration).

293: Evolution in time

294: consists in change of one dominating structures

295: by others by their enlarging

296: (the inverse cascade, changing of generations).

297: Externally

298: it reminds a situation with Rayleigh-Taylor and Kelvin-Helmholts mixing.

299: Actually mechanisms of enlarging

300: is essentially different

301: in Rayleigh-Taylor and Kelvin-Helmholts case

302: on the one hand

303: and Richtmyer-Meshkov with another.

304: 

305: 

306: 

307: 

308: 

309: In the first case

310: there is the plentiful external source

311: which amplifyes subharmonics of the current dominating scale.

312: In a Rayleigh-Taylor case it is gravitational energy,

313: reserved in a heavy liquid.

314: The heavy liquid is supported above light liquid

315: due to pressure of light liquid.

316: Therefore gravitational energy of a heavy liquid

317: cannot be liberated in free falling.

318: Downturn of the centre of gravity of system

319: from heavy and light liquids

320: occurs by exchanges or rearrangements of heavy and light liquids.

321: Accordingly Rayleigh-Taylor instability

322: is called also exchange or interchange instability.

323: In Kelvin-Helmholts case

324: a source is kinetic energy of motion

325: which is external to a turbulent mixing zone.

326: Alternation of generations

327: consists in replacement of harmonics of the current dominating scale

328: (parents)

329: on subharmonics of these harmonics

330: (descendants).

331: 

332: 

333: In the second (Richtmyer-Meshkov) case

334: an external source is not present.

335: As it will be shown in the given work,

336: fading Richtmyer-Meshkov motion

337: is supported due to statistical redistribution

338: of a longitudinal momentum from small scales in large scales.

339: Thus in the second case (as against the first)

340: energy of movement in the large scales is less

341: than energy of movement,

342: Which has been concentrated in small scales

343: on the previous stages of development of turbulent mixing.

344: 

345: 

346: 

347: In \# 2 the statement of a problem is formulated.

348: Connection between disintegration of break

349: and Richtmyer-Meshkov instability is considered

350: (generation of vorticies on contact at disintegration of break).

351: It is spoken about "processing" of surface perturbation $\eta$

352: into near boundary velocity non-uniform field $\vec v$.

353: The example of typical stochastic function

354: with the finite scale of heterogeneity $\lambda_{ini}$ is presented.

355: Functions from this class characterize general initial perturbations.

356: 

357: \noindent In \# 3 characteristics of a mixing zone are described:

358: (i) Average thickness of a mixing zone

359: in which an interpenetration of phases takes place,

360: (ii) large-scale structure (sequence of vortical dipoles),

361: and (iii) small-scale details inside large-scale dipoles.

362: 

363: 

364: \noindent In \# 4 value of the pressure

365: created in a liquid by the slowed down bubble is emphasized.

366: The important examples

367: with a periodic chain of bubbles and a solitary bubble

368: are considered.

369: 

370: 

371: \noindent In \# 5 the pressure and velocity fields

372: near a random chain of bubbles is analysed.

373: Comparison of periodic, solitary and random arrangements of bubbles

374: is instructive.

375: 

376: 

377: \noindent In \# 6 spectral characteristics of a velocity field

378: are calculated.

379: It is found,

380: that the major long wavelength part of a spectrum

381: depends on wave number as a power law.

382: 

383: \noindent In \# 7 the velocity field

384: arising directly after passage of a shock wave

385: is analysed.

386: 

387: \noindent In \# 8 the law of expansion of a turbulent mixing zone

388: into a heavy liquid is obtained

389: $$

390: (h_+\propto t^{2/5}).

391: $$

392: 

393: \noindent In \# 9 dependence on time

394: of large-scale velocity fluctuations is discussed

395: (changeability with approximately constant step on $\ln t$

396: in process of change of dominating structures in the inverse cascade).

397: 

398: \noindent In \# 10 the remark on mixing is made

399: at arbitrary value of Atwood number.

400: 

401: \noindent In \# 11 stimulated and spontaneous turbulence are compared.

402: In a stimulated case expansion of a turbulent mixing zone

403: is connected to action of beforehand created large-scale perturbations.

404: In a spontaneous case mixing develops from perturbations

405: which have finite initial scale $\sim \lambda_{ini}$.

406: As against stimulated case

407: it is a problem about decay of Richtmyer-Meshkov turbulence.

408: It is interesting to consider motion on large times

409: When thickness of a mixing layer

410: is large in comparison with initial scale,

411: $h_+\gg\lambda_{ini}$.

412: 

413: 

414: \noindent In \# 12 received results

415: are generalized on a three-dimensional case.

416: The law

417: $$

418: h_+\propto t^{1/3}

419: $$

420: of expansion of mixing front

421: is found.

422: 

423: 

424: 

425: 

426: 

427: \vspace{.5cm}

428: \centerline{{\bf \# 2. Statement of problem}}

429: 

430: \vspace{.5cm}

431: In a problem

432: about Richtmyer-Meshkov instability

433: examine the movement

434: arising owing to passage of a shock

435: through contact boundary between two different gases.

436: In fig. 1 straight lines 1 and 2

437: correspond to contact boundary (a straight line 1)

438: and a shock wave (a straight line 2).

439: 

440: At the moment $t=0$ the wave reaches boundary.

441: At $t> 0$ from boundary

442: in the different sides

443: depart or two shock waves (a case 1),

444: or a shock wave and a rarefaction (expansion) wave (a case 2)

445: (see, for example, reviews in monographies

446: Inogamov, Demianov, and Son 1999,

447: Inogamov 1999

448: and in articles Wouchuk 2001a, 2001b).

449: What from these two cases is carried out

450: depends on the density ratio,

451: thermodynamic properties of gases

452: and Mach number of a shock.

453: Usually in case of 1

454: shock goes from light gas

455: into heavy gas ("from light to heavy"),

456: and in case 2 - on the contrary.

457: Thus the rarefaction wave

458: leaves from contact boundary, moving on heavy gas.

459: 

460: 

461: 

462: 

463: Problem about a shock wave,

464: moving through plane contact boundary between two homogeneous semispaces,

465: is a special case of the well-known important problem about disintegration of break.

466: At the analysis of Richtmyer-Meshkov instability usually a case with substances,

467: still before passing of a shock, is considered.

468: Certainly, generalizations of this analysis are possible

469: which include others cases of disintegration of break

470: (generation of vorticity on and near the perturbed contact boundary

471: at disintegration of general break).

472: For example, it is interesting to consider a case

473: in which before occurrence of shock two homogeneous semispaces move towards each other

474: (Demchenko, 2004).

475: 

476: 

477: 

478: Richtmyer-Meshkov instability arises

479: when boundary between gases has perturbations.

480: In Fig. 2 the example of the perturbed boundary is shown.

481: The axis $z$ (a longitudinal axis) is directed on a normal to unperturbed boundary.

482: Below plane (2D) and three-dimensional (3D) spaces are considered.

483: The coordinate $x$ $(x, y)$ is transverse coordinate in 2D (3D) space.

484: 

485: 

486: Let's consider coordinate frame connected to unperturbed boundary.

487: After passage of a shock through boundary

488: there is a non-uniform near boundary velocity field $\vec v(\vec r, t)$.

489: It causes movings of boundary $\eta (x, y, t)$

490: concerning position of unperturbed boundary

491: (change of an amplitude of perturbations $\eta$ in time).

492: In Fig. 3 movement of boundary

493: in cases in a phase and an counterphase is shown

494: (see, for example, Inogamov, Demianov, and Son 1999, Inogamov 1999).

495: We compare here an angular phase of perturbations of boundary $\eta$

496: with a phase of velocity $\dot\eta$.

497: Movement in a phase

498: takes place at passage of a shock from light to heavy gas,

499: and in an counterphase at passage of a shock from heavy to light gas

500: (see Fig. 3

501: on which positive directions of longitudinal axes of coordinate $z$

502: and velocity $w$ are shown).

503: 

504: 

505: In Fig. 4 velocity distribution

506: $\vec v(x,z=0,t_{fix})$

507: Or $ \vec v(x, y=y_{fix}, z=0, t_{fix})$

508: along transverse coordinate $x$ (or $x, y)$ is shown

509: in 2D (or 3D) case.

510: As it was told, it is formed as a result of passage of a shock

511: through the perturbed boundary.

512: Plane $z=0$ corresponds

513: to the current position of unperturbed boundary

514: (dashed straight lines in Fig. 3).

515: Velocity components $u, v$ and $w$

516: concern to axes $x, y$ and $z$

517: (compare Fig. 3 and 2).

518: 

519: 

520: 

521: 

522: Because of the wavy form of the perturbed boundary (Fig. 2)

523: with scale $\sim\lambda_{ini}$

524: the near boundary velocity field appears sign-variable

525: with characteristic scales of wavelength $\lambda_{ini}$

526: and amplitudes $w_{\lambda ini}. $

527: Here the index "ini" specifies the initial moment of time.

528: Let's consider a case of small perturbations

529: $ \eta_{\lambda ini}^+/\lambda_{ini} \ll 1, $

530: when Richtmyer formula is applicable.

531: The scale $ \eta_{\lambda ini}^+ $

532: is a random mean square deviation of an interface

533: at shift along boundary on distance $ \sim\lambda_{ini}. $

534: The index "plus" means

535: that the given value is taken after passage of a shock.

536: As it was spoken (\# 1)

537: found below power law expansion of a turbulent mixing zone

538: remains fair for nonlinear initial perturbations (4).

539: 

540: 

541: 

542: With dimensionless factor Richtmyer formula (3) looks like

543: $$

544: w_n = F {\rm At}^+ k_n \eta_n^+ W_{unpert CB},\;\;\;

545: w_{nm} = F {\rm At}^+ k_{nm} \eta_{nm}^+ W_{unpert CB}. \eqno (5)

546: $$

547: It is used for recalculation of boundary perturbations $ \eta $

548: into velocity field

549:  $ \dot\eta $ or $w. $

550: In (5) $F $ is dimensionless factor

551: dependent in case of ideal gases from their adiabatic exponents,

552: from temperatures of gases,

553: and Mach number of a shock,

554: see, for example, monographies

555: Inogamov, Demianov, and Son 1999, Inogamov 1999

556: with reviews of articles with calculation of factor $F. $

557: In an incompressible case we have $F=1. $

558: This case corresponds to impulsive acceleration $g\propto \delta (t). $

559: In (5) At$^+ =\frac {1-\mu^+} {1 +\mu^+}, $

560: $ \mu^+ $ is the density ratio after shock,

561: $W_{unpert CB} $ is velocity of unperturbed boundary

562: (a dashed straight line, Figs. 1-3)

563: after shock

564: in laboratory frame

565: in which substances before a shock are in rest.

566: In case of small perturbations

567: at a linear stage of development of instability

568: a principle of linear superposition is valid.

569: In (5) $k_n=2\pi n/L, $

570: $n $ is number of Fourier harmonic (2D),

571: $k_{nm} = \sqrt{(k_n^x)^2 + (k_m^y)^2} $ (3D),

572: $k_{nm} $ is a wave number of a harmonic with $x $ and $y $ components

573: $k_n^x, k_m^y $ in 3D geometry.

574: We put our system into large $(L\gg \lambda_{ini})$ "box"

575: with transversal sizes $L $ (2D) Or $L\times L $ (3D).

576: Perturbations continue periodically

577: outside the end points of the "box".

578: 

579: 

580: 

581: 

582: 

583: {\bf  Comparison of amplitude of harmonic

584: and a deviation due to shift.}

585: Values $\eta_n^+$ and $\eta_{nm}^+$ are Fourier amplitudes of harmonics.

586: It is necessary to distinguish amplitudes $ \eta_n, $ $w_n $

587: of harmonics $k_n $ $ (k_n=2\pi/\lambda_n) $

588: from random-mean-square deviations due to transverse shift $ \lambda_n $

589: $$

590: \eta_{\lambda n}(t)=\sqrt{ \langle [\eta(x,t)-\eta(x+\lambda_n,t)]^2 \rangle_x },

591: $$$$

592: w_{\lambda n}(z,t)=\sqrt{ \langle [w(x,z,t)-w(x+\lambda_n,z,t)]^2 \rangle_x },\;\;\;

593: \langle f \rangle_x = \int_0^L f dx/L. \eqno (6)

594: $$

595: Equations (6) can be generalized for 3D case.

596: The random-mean-square deviation of velocity

597: depends on velocity distribution

598: which is formed by addition of many harmonics.

599: Therefore for random functions $f$

600: formed by an interference of harmonics $f_n $

601: with a random phase $ \psi_n $

602: we receive

603: $$

604: f_{\lambda n} \sim \sqrt{n} f_n. \eqno (7)

605: $$

606: 

607: 

608: 

609: 

610: 

611: \vspace{.5cm}

612: \centerline{{\bf \# 3. Structure of turbulent mixing zone}}

613: 

614: 

615: \vspace{.5cm}

616: {\bf Thickness of mixing layer.}

617: At $t <0 $ there are two semispaces "h" and "l" in contact (Fig. 5a)

618: filled with pure substances with density $ \rho_h $ and $ \rho_l. $

619: Thus a vertical (function of $z$) profile of average concentration

620: $ \bar c (z, t=0) = \langle c \rangle_{x y} $ (see (6))

621: of light phase "l"

622: is the "step"

623: which is drawn

624: by dashed straight lines in Fig. 5b.

625: At moment of time $t=0$ the shock arrives at unperturbed contact boundary.

626: After that in a picture of distribution of phases

627: a new important zone appears -

628: this is the turbulent mixing layer M (Fig. 5c),

629: filled with mixture of substances "h" and "l".

630: The quantity $c (\vec r, t) $

631: gives a volume fraction of light phase in an element of volume.

632: Density of mixture

633: which fills this volume

634: is equal

635: $$

636:  \rho_{mix} (\vec r, t) =c\rho_l + (1-c) \rho_h.

637: $$

638: After appearance of mixing zone

639: the profile $ \bar c (z, t) $ "is smeared"

640: (Fig. 5b, a continuous curve)

641: on thickness of a layer of mixing $h=h_+ + h_-. $

642: Coordinates $z=h_+ $ and $z=h_-$

643: (last points of interpenetration of phases)

644: correspond to fronts of mixing:

645: $h_+ $ is a front of bubbles, and $h_-$ is front of jets

646: $ (\rho_l <\rho_h). $

647: 

648: 

649: 

650: {\bf Transverse large-scale structures.}

651: Let us consider space structure of mixing zone.

652: The one-dimensional profile of average concentration $\bar c(z,t_{fix})$

653: (Fig. 5b)

654: reminds diffusive profiles.

655: Intermolecular distances and lengths of free paths of molecules

656: are small in comparison with thickness $h.$

657: Therefore molecular diffusive profiles are really one-dimensional.

658: In a turbulent case

659: magnitudes of the turbulent fluxes

660: carrying phases along the $z$ axis

661: and rate of expansion of a mixing zone

662: are defined by large-scale heterogeneities of mixing zone.

663: It is important

664: that the transverse sizes $\lambda_{curr}$ of these heterogeneities are large

665: $$

666: \lambda_{curr}\sim h_+. \eqno (8)

667: $$

668: 

669: 

670: 

671: 

672: 

673: 

674: Average flux $\langle c w \rangle_{x y}$ of light phase "upwards" (to $z>0$)

675: and a heavy phase "downwards" $\langle (1-c) w \rangle_{x y}$

676: are opposite.

677: The mixing zone M (Fig. 5c) is two-stream region of counter motion of phases.

678: Transport of "l" and "h" substances

679: occurs by interchangement of volumes "l" and "h":

680: h$\leftrightarrow$l (Fig. 6).

681: Usually interchanges are connected with Rayleigh-Taylor instability

682: where they free gravitational energy.

683: Here (Richtmyer-Meshkov instability)

684: interchanges (Fig. 6) occur because of inertia of motion,

685: as the only possible form of motion.

686: Certainly, any movings of liquid volume

687: to an incompressible liquid are exchange motion.

688: 

689: 

690: 

691: In space large-scale (Fig. 6) two counter stream motion

692: is carried out through alternation of jets, "fingers" or columns

693: of light "l" and heavy "h" liquids (Fig. 7a).

694: This alternation

695: causes changes of a sign on vertical velocity $w$

696: as function of transverse or horizontal coordinates (Fig. 7b).

697: Conformity of columns and signs on velocity

698: is underlined by the arrows

699: connecting Figs. 7a and 7b.

700: The correlator $K_{c w}(z,t)=\langle (c-1/2) w \rangle_{x y})$

701: is positive.

702: The current thickness $h$ of a mixing layer

703: is defined on positions of last points

704: $h_{\pm} $ (Fig. 7a).

705: Horizontal arrows show designing these points on an axis $z. $

706: 

707: 

708: 

709: Sign alternation of velocity and alternation of columns

710: are caused by a chain of vortical "spots" (Fig. 8a)

711: in which vorticity turns off.

712: Signs on circulation of "spots" in a chain alternate.

713: It is shown in Fig. 8a with the help of the bent arrows in circles

714: and in Fig. 8b by signs plus and a minus in circles.

715: Vortical pairs (dipoles)

716: involve in a zone of mixing pure "l" and "h" liquids

717: (arrows "l" and "h" in Fig. 8a).

718: 

719: 

720: 

721: 

722: Base element of large-scale structure

723: is the vortical pair or a vortical dipole (Fig. 8) of the size (8).

724: The pair the next columns is connected to each dipole.

725: Occurrence of vorticities

726: occurs because of a variation of a sign on the momentum

727: transferred by a shock to a near boundary layer of substance.

728: In turn the variation of a momentum

729: is caused by wagging about average

730: the perturbed contact boundary $ \eta^-(x, y). $

731: Vortical circulation is connected

732: to the angular momentum of liquid.

733: 

734: 

735: 

736: 

737: 

738: 

739: Vortical motion at a stage III (\# 1)

740: is described in approximation of an incompressible liquid

741: (conservation of volume).

742: Thus overflows upwards should be compensated

743: (Fig. 6)

744: by returning of equal volume of a liquid downwards:

745: $ \int_0^L \int_0^L w (x, y, z=0, t) dx dy =0. $

746: At random perturbations

747: transverse correlations between base elements (between large-scale dipoles)

748: fade on scale $\lambda_{curr}$ (8).

749: That base elements

750: appear under the order of size of one size

751: along all infinite chain

752: is caused by transversal uniformity

753: on the average

754: of initial perturbations $ \eta^-. $

755: Therefore compensation of fluxes of volume

756: $ \int\int w dx dy $

757: occurs mainly at size (8).

758: Necessity of compensation

759: results to interchanges (Fig. 6) and to alternation of columns (fig. 7).

760: 

761: 

762: {\bf Previous stages and fractal structure.}

763: The instant spatial picture is above submitted.

764: Its basis is the large-scale chain of the dipoles set in scales

765: $\lambda$ (8) è $w_{rms},$

766: where

767: $$

768: w_{rms}(t)= [ \int_0^L \int_0^L [w(x,y,z=0,t)]^2 dx dy/L^2  ]^{1/2}. \eqno (9)

769: $$

770: Amplitudes of fluctuations (9)

771: is of order of random mean square

772: $w_{\lambda}$ (6) at dipole size.

773: 

774: 

775: 

776: 

777: 

778: Let this instant picture concerns to the moment of time $t^i. $

779: The index $i =\ln (t/t_{II}) $ (2)

780: runs consecutive stages of development of the inverse cascade

781: (sequence of enlarging of scale $ \lambda^i). $

782: At each enlarging the index $i $ changes on size $ \Delta i\sim 1. $

783: In Fig. 9 two large-scale chains of dipoles $ \lambda^i $ at stages $i_1 $

784: and $i_2 $ $ (i_2-i_1\gg 1)$

785: are shown.

786: At a stage $t^{i2} $ in a dipole $a_2 $

787: there are almost all dipoles $a_1 $ from a stage $t^{i1} $

788: got in an interval $x_{a2} <x <x_{a2} + \lambda^{i2}. $

789: Thus, the dipole $a_2 $ is the claster from the specified dipoles $a_1. $

790: 

791: 

792: 

793: 

794: 

795: 

796: 

797: In Fig. 10a dipoles of two consecutive generations

798: $i_1 $ and $i_2 = i_1 + \Delta i $ are shown.

799: In Fig. 10b three generations

800: $i_1, $ $i_2 = i_1 + \Delta i $ and $i_3 = i_2 + \Delta i $

801: are shown.

802: Large-scale vortical rotation strongly deforms and stretches

803: small-scale details connected to the previous generations (Fig. 11).

804: The analysis of numerical calculations shows

805: that these details of current dominating structure

806: are more likely relicts of previous generations

807: than result of Kolmogorov (3D) or Kraichnan fragmentations.

808: In such hierarchical way

809: internal fractal structure of large-scale formations forms.

810: This fractal

811: is responsible for rate of mixing of phases at a molecular level,

812: see Redondo and Garzon, 2004.

813: 

814: 

815: 

816: 

817: 

818: After separation molecular mixture remains as an intermediate layer

819: in experiments on separation with change of a sign on acceleration,

820: Kucherenko et al., 2004a.

821: The contents of light phase

822: in unit of volume of a mixture miscellaneous (varies from 0 up to 1).

823: Studying an intermediate layer

824: it is possible to establish

825: what share of the given layer

826: having this or that contents of light phase.

827: 

828: 

829: 

830: 

831: 

832: 

833: To estimate a share of molecular mixed mixture

834: in total amount of mixture

835: in turbulent zone

836: it is possible by use of chemical indicators

837: (Linden, Redondo, Youngs, 1994;

838: Kucherenko et al., 2004b).

839: Recent work (Chertkov, 2003)

840: is devoted to studies of process of turbulent fragmentation

841: during development of Rayleigh-Taylor mixing.

842: 

843: 

844: 

845: 

846: \vspace{.5cm}

847: \centerline{{\bf \# 4. Pressures and momentum, At$=1$}}

848: 

849: 

850: \vspace{.5cm}

851: Let's describe calculation of an exponent $ \theta $

852: in a case $ \mu=0, $ At $ = 1 $ (mixing with "vacuum").

853: At a stage III flow tends to universal (self-similar) regime (\# 1)

854: in which $h_+\propto t^{\theta}. $

855: Calculation $ \theta $ is based on the law of momentum conservation.

856: Transfer of momentum to liquid

857: is carried out due to ram pressure $ \rho v^2 $

858: of bubbles "imprinting" into liquid.

859: Therefore creation of momentum by pressure is interesting for us.

860: 

861: 

862: 

863: 

864: {\bf Pressure distribution. Periodical case.}

865: Let's consider a periodic chain of bubbles.

866: Periodicity means that all bubbles in a chain are identical.

867: Bubbles are symmetric concerning the vertical average line

868: (b$\infty), $

869: see Fig. 12b.

870: The important point b is an apex of a bubble.

871: In this figure the symmetric half of period of flow is shown.

872: To receive all flow

873: it is necessary to complete symmetrically to the left

874: half of bubble

875: up to the whole bubble.

876: The full chain is obtained by periodic continuation

877: of the whole bubble to the left and to the right\footnote{

878: About periodic solutions in 2D and 3D spaces see, for example,

879: reviews Inogamov, Demianov, and Son 1999;

880: Inogamov 1999;

881: and works Inogamov and Oparin, 1999;

882: Inogamov et al., 2001a.

883: Here for simplicity the example in 2D a case is presented.}.

884: 

885: 

886: 

887: 

888: 

889: 

890: Velocities of lagrangian particles

891: on a straight line (b$\infty) $

892: are directed upwards (to $z> 0), $

893: arrows B in Fig. 12a and b,

894: "imprint" of bubbles.

895: These velocities (in the straight b$\infty) $

896: decrease in due course

897: (deceleration of bubbles, inertial imprint of bubbles).

898: Hence,

899: on the straight line (b$\infty) $

900: the vector $ \nabla p $

901: is directed upwards.

902: Distribution of pressure

903: $p (x=x_b, z, t_{fix}) $ on an axis $z $

904: is shown in Fig. 12a,

905: $x_b $ is coordinate of the bubble apex b.

906: In case of an incompressible liquid

907: the reference mark of pressure can be any function of time.

908: In region occupied by "vacuum"

909: (the region below the boundary of bubble bj, Fig. 12b)

910: pressure is homogeneous:

911: $p (\vec r, t) \equiv p_{CB} (t). $

912: Let's put

913: $p_{CB} \equiv 0 $

914: (a free choice of "calibration" of pressure).

915: Then function $p(z) $ in Fig. 12a begins from zero $p(z=z_b) =0. $

916: 

917: 

918: 

919: Function $p(z) $ in Fig. 12a

920: grows with growth $z$

921: (as it is told, $dp/dz> 0). $

922: Asymptotically

923: at the large removals from roughnesses of boundary bj

924: $ (z\gg\lambda) $

925: heterogeneity of pressure fades -

926: pressure becomes homogeneous:

927: $$

928: p\to p_{\infty}(t)> p _ {CB} =0,

929: $$

930: above $ \lambda $ is the period of a chain of bubbles (Fig. 12b).

931: It is shown in Fig. 12a and c.

932: 

933: 

934: 

935: 

936: In a liquid

937: the gradient $ \nabla p $

938: is finite on boundary bj between liquid and "vacuum".

939: In jets j (liquid deformations aside "vacuum")

940: the gradient $ |\nabla p| $ is small.

941: Vertical straight line (j$\infty) $ is a middle line of a jet.

942: In Fig. 12b the family of instant isobars 0, 1, 2 and 3 is presented.

943: Pressure accrues with growth of number of an isobar.

944: Zero isobar is the boundary bj.

945: 

946: 

947: 

948: 

949: 

950: 

951: Initial kinetic energy of motion of liquid

952: gradually passes in kinetic energy of jets.

953: The momentum

954: $$

955:  \int_0^t p_{\infty} dt 

956: $$

957: passes into a momentum of the substance

958: which have been "thrown out" in jets j.

959: Let's result an example.

960: Let the condition with plane boundary

961: and one harmonic of velocity was initial:

962: $$

963: z =\eta (x, t=0) \equiv 0, \;\;\;

964: \vec v = \nabla \varphi, \;\;

965: \varphi (x, z, t=0) = - (w_{ini}/k) e^{-kz} \cos kx,\;\;z> 0.

966: $$

967: Then initial vertical $ (w =\varphi_z) $ momentum

968:  $ \int_{-\infty}^{\infty} dx \int_0^{\infty} w dz $

969: is equal to zero.

970: At $t=0 $ velocities of apex points b

971: $ (w_b, x_b=0) $

972:  and j $ (w_j, x_j =\pi/k) $

973: are equal each other $ \pm w_{ini}. $

974: At $t\to\infty $ absolute value of velocity $w_j $

975: is approximately twice larger than $w_{ini},$

976: Inogamov et al., 2001a.

977: Integral

978: $$

979: I(t) =

980: \lim_{L\to\infty}

981: | \int_0^L (dx/L) \int_{\eta (x, t)}^{\infty} w dz |

982: $$

983: (where function $ \eta (x, t) $

984: gives the current position of contact boundary)

985: accrues in due course $t $

986: and tends to limiting value

987: $$

988:  \int_0^{\infty} p_{\infty}/\rho_h dt. 

989: $$

990: 

991: 

992: 

993: So, first, in a periodic case we have

994: $p_{\infty}> p_{CB}$

995: (the case $ \mu=0$ is considered),

996: $$

997: p_{\infty}-p_{CB}\sim \rho_h w_b^2. \eqno (10)

998: $$

999: Second, heterogeneity of pressure are connected to roughnesses of boundary.

1000: 

1001: 

1002: 

1003: 

1004: 

1005: 

1006: 

1007: Let's note that at $ \mu\neq 0 $ "hole" or depression in pressure is connected

1008: with compact region of vortical mixing

1009: (pressure in vortical region is lower than $p_{\infty}). $

1010: Let's consider mixing near a horizontal plane $z=0 $ (3D)

1011: or a horizontal straight line $z=0 $ (2D).

1012: When plane mixing zone is present (Fig. 13a)

1013: then in distribution of pressure "valley" (2D)

1014: extended along an axis $x$ is formed.

1015: Average $ \langle\rangle_x $ (6)

1016: instant profile of pressure $p(z) $ is shown in Fig. 13b.

1017: Differences of pressure

1018: $\Delta p_{\infty}^h = p_{\infty}^h - p_{min}\sim \rho_h w_b^2$

1019: and $\Delta p_{\infty}^l\sim \rho_l w_j^2$

1020: correspond to two semispaces $z\to\pm\infty $

1021: filled still pure "h" and "l" liquids

1022: with densities $ \rho_h, $ $ \rho_l, $

1023: $p_{min} $ is pressure in a minimum.

1024: At a stage $t\sim t_{II} $ (2) we have $w_b\sim w_j\sim w_{ini}. $

1025: Accordingly $ \Delta p_{\infty}^h> \Delta p_{\infty}^l $

1026: as $ \mu <1, $ $ \mu =\rho_l/\rho_h. $

1027: 

1028: 

1029: 

1030: 

1031: 

1032: 

1033: {\bf Pressure distribution. Solitary case.}

1034: Let's consider a case

1035: when into liquid semispace $ (z> \eta) $

1036: the single or solitary bubble (Fig. 14a)

1037: penetrates,

1038: $ \eta $ is contact boundary

1039: between liquid and "vacuum" bj$\infty. $

1040: In Fig. 14a the right half of a bubble is drawn.

1041: This bubble is symmetric

1042: at transformation of a horizontal axis $x\to-x. $

1043: Flow is symmetric

1044: concerning a middle vertical straight line (b$\infty)$,

1045: b is the apex of the bubble.

1046: Penetration of a bubble

1047: is accompanied by "squeezing" of liquid jets j.

1048: On a vertical straight line (b$\infty)$

1049: velocity

1050: (arrows $w $ in Fig. 14b)

1051: is positive

1052: while

1053: acceleration of Lagrangian particles

1054: (arrows "acc", "dec")

1055: changes a sign:

1056: particles which are near to the bubble are decelerated,

1057: and distant - are accelerated.

1058: 

1059: 

1060: 

1061: 

1062: 

1063: 

1064: The profile of pressure with a maximum is shown in Fig. 14b.

1065: We have put $p_b=p_{CB} =0$

1066: Where $p_b $ is pressure in the apex b.

1067: On infinity pressure in a liquid tends to the pressure in "vacuum"

1068: $p_{CB}. $

1069: The system of isobars 0, 1, 2, 3, 4 is represented in Fig. 14c.

1070: The zero isobar 0 is the contact bj$\infty. $

1071: Pressure grows with growth of number of an isobar.

1072: 

1073: 

1074: 

1075: 

1076: 

1077: 

1078: 

1079: 

1080: 

1081: \vspace{.5cm}

1082: {\bf \# 5. Pressure and velocity distributions in a random case}

1083: 

1084: 

1085: 

1086: \vspace{.5cm}

1087: Let's consider a random chain (Fig. 15).

1088: Useful example of a random chain

1089: (together with numerical calculation its correlation

1090: and spectral characteristics)

1091: is presented in work Inogamov et al., 2001b.

1092: Horizontally homogeneous chains

1093: (which are made from comparable bubbles)

1094: are the most interesting.

1095: Let $ \lambda_{ini} $ is the average bubble size.

1096: The individual sizes of bubbles $ \lambda_i $ (Fig. 15)

1097: are randomly scattered about average value $ \lambda_{ini} $

1098: with a dispersion $ \Delta\lambda\sim \lambda_{ini}$.

1099: Let's speak

1100: "periodicity about randomness":

1101: $Q =\lambda_{ini}/\Delta\lambda\sim 1$

1102: where $Q $ is good quality (Q-factor).

1103: 

1104: Penetration of horizontally unbound bubble chains

1105: raises pressure upon infinity $p_{\infty} $

1106: in comparison with pressure upon boundary $p_{CB}. $

1107: Pressure $p_{\infty} $ is created

1108: by ensemble of bubbles

1109: which is operating as the whole.

1110: This pressure

1111: is a measure of an average ram pressure $ \rho v^2 $

1112: of bubbles in a chain.

1113: Concerning squeezing of infinitely removed liquid

1114: the action of the random chain is similar with periodic chain (\# 4).

1115: 

1116: 

1117: 

1118: 

1119: 

1120: 

1121: 

1122: 

1123: 

1124: In case of identical bubbles

1125: (a periodic chain)

1126: pressure in liquid is less than $p_{\infty} $

1127: (Fig. 12, if $z<\infty$)

1128: and heterogeneity of pressure is horizontally periodical.

1129: The pressure created by solitary bubble

1130: is more than $p_{\infty} $ (Fig. 14).

1131: Movement of random ensemble

1132: creates a pressure field $p(\vec r, t_{fix}) $

1133: with a limit $p_{\infty}(t_{fix}) $

1134: and random fluctuations

1135: in which pressure can be both

1136: or higher, or lower than limiting value $p_{\infty}$.

1137: 

1138: 

1139: 

1140: 

1141: 

1142: Important

1143: that fluctuations $p$

1144: form the hierarchical structure

1145: penetrating deeply into bulk of liquid in space.

1146: It is connected to formation of statistical groups of bubbles

1147: with different size of a ram pressure, average for the given group.

1148: 

1149: 

1150: 

1151: 

1152: 

1153: 

1154: 

1155: 

1156: {\bf $"\lambda$ groups" and $"\lambda$ fluctuations" of pressure.}

1157: Let's describe

1158: hierarchical structure of heterogeneity of an instant field of pressure.

1159: Let's suppose

1160: that the ensemble is placed in "box"

1161: of the size $L\gg\lambda_{ini}$

1162: with periodic boundary conditions at the box edges.

1163: Let's break length $L $ into equal pieces

1164:  $ \lambda\gg\lambda _ {ini} $

1165: containing many bubbles $ ("\lambda$-group" of bubbles).

1166: Groups of bubbles on pieces $ \lambda $ are various (Fig. 15).

1167: Influence of bubble group

1168: from a piece $ \lambda $

1169: reaches a depth of the order $ \lambda $

1170: from boundary into liquid.

1171: 

1172: 

1173: 

1174: 

1175: 

1176: 

1177: In Fig. 16a

1178: the figure 1

1179: marks average position of contact boundary ("vacuum" is below).

1180: The length of the pieces 3 is proportional to pressure

1181: $p_{\lambda}$

1182: in the center of a square $ \lambda\times\lambda $

1183: Worth the bottom side on the boundary.

1184: Changeability of this pressure

1185: is caused by dissimilarity of $"\lambda$-groups" of bubbles.

1186: The braces 2

1187: show average pressure $ \langle p\rangle_x $

1188: on distance $ \lambda/2 $ from the boundary.

1189: 

1190: 

1191: 

1192: 

1193: 

1194: 

1195: 

1196: Creation of momentum of liquid in square cells

1197: $ \lambda\times\lambda $

1198: is connected to fluctuations of pressure

1199: $$

1200: \Delta p_{\lambda} = p_{\lambda} - \langle p\rangle_x, \eqno (11)

1201: $$

1202: as the homogeneous component $ \langle p\rangle_x $ does not make movement.

1203: Fluctuations of pressure (11) exponentially fade

1204: at distances from boundary exceeding $ \lambda$.

1205: Therefore at an estimation of momentum

1206: it is necessary to take cells with height of the order $ \lambda. $

1207: 

1208: 

1209: 

1210: 

1211: 

1212: Let's write down an estimation of momentum

1213: $$

1214: \lambda^2 \rho w_{\lambda} \sim

1215: (\rho w_{\lambda ini}^2) \lambda_{ini}

1216: \sqrt{\lambda/\lambda_{ini}} \, t_{ini}. \eqno (12)

1217: $$

1218: In (12) $ \lambda^2$ is the area of a square,

1219: and $w_{\lambda}$

1220: is an increment of velocity of substance in this square

1221: due to pressure action.

1222: At the left side of equation (12)

1223: there is a change of momentum in time $t_{ini}$

1224: per unit length in a transverse direction.

1225: On the right side

1226: in parentheses the average ram pressure of bubble is written down.

1227: Product of this pressure on $\lambda_{ini}$

1228: gives average force per unit of transverse length.

1229: With such force bubble presses on liquid.

1230: Statistical multiplier

1231: as a root from an average number of bubbles

1232: in a $"\lambda$-group" $N_{\lambda}$

1233: gives estimate of random fluctuations of pressure (11),

1234: $N_{\lambda} = \lambda/\lambda_{ini}. $

1235: The last multiplier on the right side of equation (12)

1236: is time of existence of bubble generation of scale $\lambda_{ini}. $

1237: 

1238: 

1239: 

1240: 

1241: 

1242: 

1243: 

1244: As we see,

1245: momentum bubble groups of scale $\lambda$

1246: written down in the right part of balance (12)

1247: diverges at large $\lambda$

1248: because of the statistical factor $\propto \sqrt{\lambda}. $

1249: It is caused by statistical accumulation of momentum

1250: with growth of the size of $"\lambda$-groups".

1251: Thus velocity of motion

1252: $w_{\lambda} \propto I_{\lambda}/S_{\lambda}$ decreases,

1253: here $I_{\lambda}$ and $S_{\lambda}$ are momentum

1254: and the area of a $"\lambda$-group".

1255: 

1256: 

1257: 

1258: 

1259: 

1260: 

1261: The sign of velocity fluctuations $w_{\lambda} $ (12)

1262: randomly varies

1263: from one square $\lambda\times\lambda$ to another (Fig. 16b).

1264: Signs $w_{\lambda}$ and $\Delta p_{\lambda}$ (11) coincide.

1265: 

1266: 

1267: 

1268: 

1269: {\bf Hierarchy of $"\lambda$-groups" and dependence from $z$.}

1270: Above we have considered

1271: one near boundary layer

1272: of thickness $\sim\lambda$

1273: in which the random velocity field $w_{\lambda}$

1274: is created by $"\lambda$-groups" from $\approx N_{\lambda}$ bubbles.

1275: Horizontal (random changeability)

1276: and vertical (exponential fade) scales

1277: of spatial heterogeneity of this field

1278: is of order of the length of pieces $\lambda$, Figs. 15 and 16b.

1279: 

1280: 

1281: 

1282: 

1283: 

1284: 

1285: To find velocity field from a random chain (Fig. 15)

1286: it is necessary to consider set of splittings into pieces $\lambda$

1287: of different scale.

1288: The hierarchy of splittings

1289: (or hierarchy of $"\lambda$-groups")

1290: looks like: $ \lambda_0, \lambda_1, \lambda_2..., $

1291: Where

1292: $$

1293: \lambda_n=a^n \lambda_{ini}, \eqno (13)

1294: $$

1295: $a> 1, a\sim 1, $ for example, $a=2. $

1296: With each scale

1297: (set by the index $n$)

1298: particular near boundary layer is connected.

1299: On the one hand

1300: scales with different $n$ are independent

1301: (there are not correlations between them),

1302: on the other hand the final velocity field $w(z)$

1303: (from a chain)

1304: develops of imposing all layers $n.$

1305: 

1306: 

1307: 

1308: 

1309: For an example we shall consider

1310: $"\lambda$-groups" from ten (on the average) bubbles

1311: $(N_{\lambda} = \lambda/\lambda_{ini} =10). $

1312: Changeability on coordinate $x$ values $\Delta p_{\lambda} $ (11)

1313: and $w_{\lambda} \propto \Delta p_{\lambda} $ (12)

1314: is generated by non-equivalence of neighboring tens, Fig. 16a and b.

1315: This changeability fades on depth $\sim 10\lambda_{ini}. $

1316: According to the formula (12),

1317: the amplitude of the given changeability makes

1318: $w_{10} \propto 1/10^{3/2}. $

1319: 

1320: 

1321: 

1322: 

1323: 

1324: 

1325: 

1326: 

1327: 

1328: Let's compare $"\lambda$-groups" from tens and hundreds bubbles.

1329: The amplitude of changeability of velocity

1330: at shifts on coordinate $x$ on $100\lambda_{ini}$

1331: is less $(w_{100} \propto 1/100^{3/2}). $

1332: Therefore on depth $ \sim 10\lambda_{ini} $

1333: fluctuations $w_{100} $

1334: are imperceptible on a background of fluctuations $w_{10} $

1335: (closer over contact is dominated with fineer splittings).

1336: But on distances from boundary $ \sim 100\lambda_{ini}$

1337: fluctuations $w_{10} $ exponentially decrease,

1338: and to dominate begin fluctuations $w_{100}. $

1339: 

1340: 

1341: 

1342: 

1343: 

1344: Told it is illustrated in Fig. 17

1345: on which the spatial map of vertical velocity $w$

1346: on a plane of movement $(x, z)$ is shown.

1347: Three consecutive scales $n-1, n, n+1 $

1348: hierarchies (13) (small, average and big squares) are represented.

1349: Velocities $w_{n-1}, w_n $ and $w_{n+1}$

1350: dominate over different distances.

1351: To negative velocities there correspond the shaded squares.

1352: Thus, the field $w (x, z)$

1353: looks like randomly branching tree

1354: with a trunk $L,$

1355: sharing on $a $ (13) branches

1356: which in turn share for $a$ branches everyone, etc.

1357: 

1358: 

1359: 

1360: 

1361: 

1362: Dependence of amplitudes of fluctuations of velocity

1363: on depth $w(z) $

1364: is formed as result of "interference" of splittings $ \lambda_n. $

1365: Total action of hierarchy of splittings (13) can be written down as

1366: $$

1367: w(z)=\sum_n w_n \exp \left( - \frac{2\pi z}{\lambda_n}\right) \propto

1368: \sum_n \lambda_n^{-3/2}

1369: \exp \left( - \frac{2\pi z}{\lambda_n} \right)

1370: \propto \frac{1}{z^{3/2}},\;\;\;

1371: w(z)\sim w_{\lambda ini} \frac{\lambda_{ini}^{3/2}}{z^{3/2}}. \eqno (14)

1372: $$

1373: 

1374: 

1375: 

1376: 

1377: Step-type behaviour

1378: or lacunarity (13) (lacuna or gap) for hierarchy

1379: was necessary for simplification of explanations.

1380: It is possible

1381: to consider an interference of splittings

1382: with fine step of a geometrical progression

1383: $a=1 +\epsilon, $ $ \epsilon\ll 1. $

1384: Velocity field is smooth;

1385: its particular shape

1386: is defined by arrangement of bubbles in a chain (Fig. 15),

1387: instead of hierarchy (13).

1388: Thus of "stain" or "spots" of negative velocity $w$

1389: have smooth contours (not squares, as in Fig. 17).

1390: In Fig. 17

1391: the diagram of splitting of consecutive scales

1392: of these "spots", instead of their form is submitted.

1393: 

1394: 

1395: 

1396: 

1397: 

1398: The Fig. 18 supplements the formula (14).

1399: Points $n-1, n $ and $n+1 $

1400: mark the centers of squares, compare Figs. 17 and 18a.

1401: These centers are on distances $ \lambda_n/2 $

1402: from the boundary submitted by a horizontal piece.

1403: The specified points are transferred on a plane $ (w, z) $

1404: (Fig. 18b)

1405: with the help of three horizontal arrows

1406: connecting the left and right parts Fig. 18.

1407: Coordinates of these points are

1408: $ (w_n\propto 1/\lambda_n^{3/2}, z_n =\lambda_n/2). $

1409: Through sequence of the transferred points

1410: the curve $w(z) $ (14) is drawn.

1411: As we see,

1412: small-scale fluctuations $w_{\lambda} $ (12)

1413: have the big amplitude

1414: whereas large-scale will penetrate deeply into thickness of liquid.

1415: 

1416: 

1417: 

1418: 

1419: Now it is possible to compare instant velocity fields $w (x, z) $

1420: from a periodic chain (\# 4),

1421: from a solitary bubble (\# 4),

1422: and from a random chain (\# 5).

1423: These

1424: qualitatively various three cases

1425: definitely supplement each other.

1426: In the first case decay on $z $ occurs exponentially.

1427: Power law dependence from $z$

1428: is connected to a solitary (or lonely) bubble,

1429: however distribution of velocity on coordinate $x $ is non-uniform.

1430: Velocity field from a random chain,

1431: first,

1432: is homogeneous on transverse distances

1433: large in comparison with the size of a bubble,

1434: and, second,

1435: as a power law depends on coordinate $z. $

1436: As it will be visible below,

1437: power law dependence $w $ from $z $

1438: results in the power law of expansion

1439: of a turbulent mixing zone

1440: $(h\propto t^{\theta}). $

1441: In a periodic case mixing quickly fades

1442: (displacement of bubbles $h\propto \ln t). $

1443: 

1444: 

1445: 

1446: \vspace{.5cm}

1447: \centerline{{\bf \# 6. Spectrum}}

1448: 

1449: 

1450: \vspace{.5cm}

1451: Let's calculate Fourier spectrum

1452: of an instant velocity field $w (x, z=0) $ created by a random chain.

1453: The scale

1454: $$

1455:  \lambda_{curr},\;\; k_{curr} =2\pi/\lambda_{curr},\;\;

1456:  n_{curr} =L/\lambda_{curr}

1457: $$

1458: divides a spectrum on long-wave

1459:  $ (n <n_{curr}) $

1460: and short-wave $ (n> n_{curr}) $ parts.

1461: The size $ \lambda_{curr} $

1462: corresponds to scale of dominating structure

1463: of the current generation of large-scale vorticies or the dipoles

1464: developed by the moment of time $t\sim t_{curr},$

1465: see Fig. 7, 8 and \# 3.

1466: For questions examined in the given work

1467: the short-wave part is unimportant.

1468: The long-wave part

1469: is determined by fluctuations $w_{\lambda} $ (12)

1470: at $ \lambda> \lambda_{curr}. $

1471: 

1472: 

1473: 

1474: 

1475: 

1476: 

1477: 

1478: 

1479: The scale of change of velocity $w_{\lambda} $

1480: at length $ \lambda, $ see (6),

1481: is connected to amplitudes of harmonics $w_n $ by equation (7).

1482: Let's write down

1483: $w_{\lambda} \sim \sqrt{n} w_n, $

1484: where numbers of harmonics $n $ correspond to scale $ \lambda: $

1485: $n\sim n_{\lambda} =L/\lambda. $

1486: From here we receive

1487: $$

1488: w_n = w_{\lambda ini} \, (\lambda_{ini}/L)^{3/2} \,

1489: \xi_n \cos\psi_n \, n, \eqno (15)

1490: $$

1491: where $ \xi_n\sim 1 $ is dimensionless number

1492: and $ \psi_n $ is the random phase.

1493: It turns out,

1494: that a long-wave spectrum is linear under number $n. $

1495: At $t\sim t_{curr} $ the formula (15)

1496: is valid up to $n <n_{curr}$\footnote{

1497: Instead of for $n <n_{ini} =L/\lambda_{ini}. $

1498: In time $t\sim t_{curr} $

1499: the maximum of a spectrum

1500: is displaced from $n\sim n_{ini}$ in $n\sim n_{curr}. $

1501: }.

1502: At a derivation was considered,

1503: that velocity field $w (x, z, t\sim t_{ini}) $

1504: keeps the kind (under the order of size)

1505: in scales $ \lambda> \lambda_{curr} $

1506: down to the moments of time $t\sim t_{curr}$\footnote {

1507: To discussion of this question are devoted \# 8 and \# 9 below.}

1508: 

1509: 

1510: 

1511: 

1512: \vspace{.5cm}

1513: \centerline{{\bf \# 7. Random chain and a shock}}

1514: 

1515: 

1516: 

1517: \vspace{.5cm}

1518: We research velocity field $w^+(x, z)$

1519: formed by a shock during short sound time $t_s $ (1)

1520: near a random chain of linear perturbations $\eta^-$ (\# 1, \# 2).

1521: Equation (5) connects harmonics of function $ \eta^+(x) $

1522: and harmonics of initial velocity\footnote{

1523: Initial for a stage of movement in a mode of an incompressible liquid.

1524: }

1525: Sequence of three steps of calculation following.

1526: First, we define the Fourier spectrum $\eta^+_n$

1527: of random surface perturbations $\eta^+(x)$\footnote{

1528: It appears,

1529: the long-wave part of a spectrum $ \eta_n $

1530: does not depend from $n, $ $ \eta_n\propto n^0. $

1531: }.

1532: Second, under Richtmyer formula

1533: the amplitude of harmonics $w_n$ is defined.

1534: Thirdly, on a spectrum $w_n $ is function $w^+ (x, z)$ defined

1535: and dependences of this function from $x$ and $z$ are analyzed.

1536: Corresponding calculations are executed in work Inogamov, 2002.

1537: It is shown, first,

1538: similarly (15) spectrum $w_n $ is linear on $n,$

1539: second, dependence $w $ from $z $ is given by the formula (14).

1540: 

1541: 

1542: 

1543: \vspace{.5cm}

1544: {\bf \# 8. Decay of velocity with depth and expansion law}

1545: 

1546: 

1547: \vspace{.5cm}

1548: Let's address to a question on evolution of velocity field $w (x, z, t) $

1549: (the given section and the following).

1550: After a push connected to a shock wave,

1551: there is velocity field $w^+(x, z) $ (\# 7).

1552: We admit, in the further field $w^+$ remains "frozen" in liquid.

1553: In the following section we shall return to this assumption.

1554: Let's explain, that means "frozen".

1555: Basis is that

1556: (as it was spoken \# 5)

1557: small-scale modes have large velocities,

1558: and large-scale will penetrate deeply into liquid.

1559: 

1560: 

1561: The field of velocity $w^+(x, z)$

1562: with enlarging cells

1563: (Figs. 17, 18a)

1564: and the power law decay on $z$ (Fig. 18b) is formed.

1565: 

1566: 

1567: 

1568: 

1569: 

1570: 

1571: 

1572: 

1573: 

1574: At such position the assumption about "frozen" cells is natural.

1575: Really,

1576: turning time of large formations is large

1577: $$

1578: t_{\lambda} \sim \lambda/w_{\lambda} \propto t^{5/2}. \eqno (16)

1579: $$

1580: Change of velocity $w_{\lambda}$

1581: on depth $z\sim\lambda$

1582: occurs {\it after penetration of contact boundary

1583: on this depth $h_+\sim\lambda.$}

1584: If it so it is possible to write down

1585: $$

1586: \frac{dh_+}{dt} \sim w(z)|_{z\sim h_+} \sim 1/h_+^{3/2}.

1587: $$

1588: From here follows the required law\footnote{

1589: Several phenomenological approaches

1590: was earlier offered

1591: (K-model and buoyancy-drag model)

1592: for the approximate estimation of an exponent $ \theta.$

1593: Usually

1594: reduction of this exponent

1595: in comparison with value 2/3 (conservation of energy)

1596: is connected with viscous losses of energy on small scales.

1597: The review of these approaches is given in work Inogamov 2002.

1598: 

1599: In work of Schwarz et al., 2000

1600: the exponent $\theta$ for 2D and 3D cases

1601: has been extimated with the help of bubble envelop model.

1602: The envelop model

1603: essentially distinguishes

1604: from K-model and buoyancy-drag model.

1605: Values $\theta_{2D} \approx 0.4$

1606: and    $\theta_{3D} \approx 0.2$

1607: have been received by the envelop model (Schwarz et al., 2000).

1608: 

1609: As against the listed approaches

1610: in the present work

1611: the exponents $\theta_{2D}$ (17)

1612: and $\theta_{3D}$ (\# 12 below)

1613: are calculated precisely.

1614: The mechanism of expansion of mixing zone is opened.

1615: 

1616: Appeared,

1617: dynamics of expansion

1618: is defined by long wavelength $ (\lambda> \lambda_{curr})$

1619: fluctuations of velocity spectrum.

1620: At the moment of time $t_{curr}$

1621: these fluctuations

1622: are at a linear stage of development

1623: (displacement of liquid particles $\eta_{\lambda}$

1624: on scale $\lambda$

1625: are small in comparison with $\lambda).$

1626: }

1627: of decay of Richtmyer-Meshkov turbulence

1628: $$

1629: h_+\sim w_{\lambda ini}^{2/5} \,\lambda_{ini}^{3/5}\, t^{\theta},

1630:  \; \; \; \; \;

1631: \theta = 2/5. \eqno (17)

1632: $$

1633: Large-scale formations

1634: enter game one by one

1635: in the hierarchical order (13)

1636: and with the big delay (16).

1637: This gradual inclusion of more and more large cells

1638: defines the law $h_+(t)$ (17).

1639: 

1640: 

1641: 

1642: 

1643: 

1644: \vspace{.5cm}

1645: {\bf \# 9. Accumulation of momentum and logarithmic corrections}

1646: 

1647: 

1648: 

1649: \vspace{.5cm}

1650: Velocity field $w^+(x, z) $

1651: is formed (\# 7)

1652: for a time interval $\sim t_s $ (1).

1653: Further in time

1654: $\sim t_{\lambda ini}$ (2)

1655: "balls" (Fig. 8) primary vorticies

1656: are turned off -

1657: the dominating structure of the first generation is formed.

1658: At $t\sim t_s$ displacement of contact boundary

1659: are small (still there is no turning boundary).

1660: The random chain of sign-variable vorticies

1661: $\lambda_{ini}, w_{\lambda ini}$

1662: forms velocity field $w(x, z, t\sim t_{\lambda ini})$

1663: in near boundary layer.

1664: Important that fields $w^+(x, z)$ and $w(x, z, t_{\lambda ini})$

1665: under the order of size are identical.

1666: 

1667: 

1668: 

1669: 

1670: 

1671: 

1672: 

1673: Further process of change of one generations by the following develops.

1674: Let's consider the current generation

1675: with scale of dominating structure $\lambda_{curr}.$

1676: The random chain $\lambda_{curr}, w_{\lambda curr}$

1677: forms velocity field

1678: $$

1679: w(x, z, t\sim t_{\lambda curr}) \sim

1680: w_{\lambda}(t_{\lambda curr}) \sim

1681: w_{\lambda curr} (\lambda_{curr}/\lambda)^{3/2} \eqno (18)

1682: $$

1683: in surpassing scales $ \lambda> \lambda_{curr}.$

1684: Substituting in (18)

1685: following from the law (17) estimations

1686: $$

1687: \lambda_{curr} \sim h_+(t_{\lambda curr}),

1688:  \; \; \; 

1689: w_{\lambda curr} \sim

1690:  \frac{dh_+(t_{\lambda curr})}{dt},

1691: $$

1692: we express velocities $w(x, z, t\sim t_{\lambda curr})$

1693: or $w_{\lambda}(t_{\lambda curr})$ in (18)

1694: through parameters of primary vorticies $\lambda_{ini}, w_{\lambda ini}.$

1695: It turns out,

1696: that in the rest by the moment $t\sim t_{\lambda curr}$

1697: undeveloped (or "unsaturated")\footnote{

1698: The latent or "frozen" scales.

1699: }

1700: scales $\lambda> \lambda_{curr}$

1701: all generations of dominating structures

1702: create under the order of size the same field of velocity $\sim w^+!$

1703: 

1704: 

1705: 

1706: 

1707: 

1708: At discussion of velocities $w(x, z, t_{\lambda curr})$

1709: or $w_{\lambda}(t_{\lambda curr})$

1710: the question was amplitudes of fluctuations.

1711: Let's consider now the phase information.

1712: On amplitude velocities

1713: created by different generations

1714: are of the same order of magnitude value.

1715: It is asked,

1716: whether the fields

1717: created by different generations

1718: are identical on a phase?

1719: In other words,

1720: whether the shaded cells on a map of velocities

1721: from different generations (Fig. 17) coincide?

1722: 

1723: 

1724: 

1725: 

1726: There are three opportunities.

1727: First,

1728: the large-scale part $ (\lambda> \lambda_{curr})$

1729: fields $w(x, z) $

1730: can practically not vary in time.

1731: In this case the law of mixing (17) is carried out.

1732: Second,

1733: structure of shaded cells

1734: formed by different generations

1735: can coincide under the order of magnitude.

1736: Thirdly,

1737: this structure can randomly vary

1738: from generation to generation.

1739: 

1740: 

1741: 

1742: 

1743: 

1744: 

1745: 

1746: Let's consider last two opportunities.

1747: Let's recollect,

1748: that $w_{\lambda}(t_{\lambda curr}) $

1749: is {\it an increment} of velocity

1750: connected to an increment of momentum

1751: for a time interval $\sim t_{curr}. $

1752: Means for the finding of velocity field

1753: developing by the moment of time $t\sim t_{curr}$

1754: it is necessary to summarize all increments

1755: caused by activity of the previous generations.

1756: 

1757: 

1758: 

1759: 

1760: In case the second opportunity is realized, we receive

1761: $$

1762: w_{\lambda}(t_{\lambda curr}) \sim

1763: \ln (\lambda_{curr}/\lambda_{ini}) \, w^+. \eqno (19)

1764: $$

1765: In the third case we have

1766: $$

1767: w_{\lambda}(t_{\lambda curr}) \sim

1768: \sqrt {\ln (\lambda_{curr}/\lambda_{ini})} \, w^+. \eqno (20)

1769: $$

1770: Accumulation of increments

1771: results in logarithmic amplification.

1772: The logarithmic corrections

1773: are connected with a little bit strengthening of mixing.

1774: In view of it instead of the law (17) it is received

1775: $$

1776: h_+\propto (\ln t)^{2/5} \; t^{2/5}, \; \; \; \; \;

1777: h_+\propto (\ln t)^{1/5} \; t^{5/2}

1778: $$

1779: in cases (19) and (20) accordingly.

1780: 

1781: In 3D case we have

1782: $$

1783: h_+\propto (\ln t)^{1/3} \; t^{1/3}, \; \; \; \; \;

1784: h_+\propto (\ln t)^{1/6} \; t^{1/3}.

1785: $$

1786: 

1787: 

1788: 

1789: 

1790: \vspace{.5cm}

1791: \centerline{{\bf \# 10. Influence of inertia of light phase}}

1792: 

1793: 

1794: \vspace{.5cm}

1795: In the previous sections

1796: the case $\mu=0$

1797: (At $=1,$ contact to "vacuum", $\mu$ is density ratio)

1798: was investigated.

1799: The case $ \mu\neq 0 $

1800: (the account of density of light phase)

1801: is difficult, and demands separate consideration.

1802: Position at $ \mu\neq 0$ in 2D geometry is difficult.

1803: In this case kinetic energy of an incompressible liquid conserves\footnote{

1804: Let's note, that in a case $\mu=0$ viscous dissipation

1805: at Re$\to\infty $ is absent in 3D geometry.

1806: }

1807: for the case of small viscosity (Re $\gg 1).$

1808: In a case $\mu=0$

1809: the viscous heat production can be neglected

1810: at large values of Reynolds number Re

1811: (conservation of kinetic energy).

1812: However

1813: difficulty with excess of energy in small scales\footnote{

1814: Formally

1815: the law of conservation of energy

1816: results in value $ \theta=2/3$

1817: which essentially exceeds the value (17).

1818: }

1819: at $\mu=0$ does not arise.

1820: The matter is that

1821: the fly of fast jets in "vacuum" carries away surplus of energy.

1822: At such position

1823: promotion of front $h_+ $

1824: aside heavy liquid

1825: is defined

1826: by the law of conservation of momentum

1827: which results in the formula (17).

1828: 

1829: 

1830: 

1831: Flow with the power law of expansion $h_+\propto t^{\theta}$

1832: is self-similar.

1833: Thus in self-similar variables

1834: $$

1835: t^{\theta}\, n,\; \; \, t^{1-3\theta/2}\, w_n 

1836: $$

1837: spectrum (\# 6) is steady-state\footnote{

1838: It is steady-state on the average.

1839: Thus there are fluctuations of amplitudes

1840: about average values and fluctuations of phases.}.

1841: The degree scaling amplitudes,

1842: appears from an equation:

1843: $ \sqrt{n}\, w_n\sim

1844:  \dot h_+$ at $n\sim n_{curr},$ see (7)

1845: and \# 6 (definition of $n_{curr}). $

1846: 

1847: 

1848: 

1849: In self-similar flow

1850: the profile of average density

1851: $\bar\rho = \langle\rho\rangle_{\bot}$

1852: is function not of two $(z, t)$

1853: but of one variable\footnote{

1854: Certainly,

1855: self-similar asymptotics

1856: is established in process of growth of number of the generations

1857: replaced

1858: after generation of primary vorticies

1859: $\lambda_{ini}, w_{\lambda ini}.$

1860: },

1861: $\bar \rho(z/h_+)$

1862: or $\bar \rho (z/t^{\theta});$

1863: the mark $\bot$ means averaging on transverse coordinates.

1864: The coordinate $z$

1865: is counted from initial position of boundary (\# 1, \# 2).

1866: At $ \mu=0 $

1867: function $ \bar\rho $ quickly decreases aside "vacuum".

1868: It becomes very small

1869: already at value of dimensionless coordinate

1870: $(-z)/h_+(t)$

1871: about several units.

1872: 

1873: 

1874: 

1875: 

1876: First portions of the substance

1877: (which have been thrown out by primary generation of bubbles)

1878: flying ahead aside "vacuum"

1879: move under the law $h_-\sim - w_{\lambda ini} t.$

1880: Distribution of density near to front $h_-$

1881: is not described by the self-similar solution $ \bar \rho (z/h_+).$

1882: 

1883: 

1884: 

1885: 

1886: 

1887: \vspace{.5cm}

1888: \centerline{{\bf \# 11. Stimulated and spontaneous turbulence}}

1889: 

1890: 

1891: 

1892: \vspace{.5cm}

1893: {\bf Addition of initial long wavelength spectrum.}

1894: Problem I is above considered

1895: in which the long wavelength part

1896: $(\lambda> \lambda_{curr})$

1897: of a spectrum $w_n $ or $w_{\lambda}$

1898: is formed owing to an interference of random small-scale modes.

1899: Let's address to other problem

1900: (a problem II)\footnote{

1901: Below problems I and II refer to

1902: as problems about stimulated and spontaneous turbulence.

1903: }.

1904: In it

1905: to a random initial chain of surface perturbations

1906: $\lambda_{ini}, \eta_{\lambda ini}$

1907: the long wavelength spectrum is added

1908: $$

1909: \eta^{LW}_n =

1910: \frac{

1911: \lambda_{\star}^{\beta+3/2}

1912: }

1913: {L^{\beta+1/2}}\,

1914: \xi_n\, \cos\psi_n\; n^{\beta},\;\;\;\;\;

1915: 1\leq n\leq n_{ini}=L/\lambda_{ini},

1916: \eqno (21)

1917: $$

1918: where $\langle\xi_n\rangle =1,$

1919: $\psi_n $ is a random phase,

1920: $L $ is length of "box" (\# 2).

1921: Let's write down also

1922: $$

1923: \eta^{LW}_{\lambda}=

1924: (\lambda_{\star}^{\beta+3/2}/L^{\beta+1/2})\, n^{\beta+1/2},\;\;

1925: \epsilon^{LW}_{\lambda}=\eta^{LW}_{\lambda}/\lambda\sim

1926: (\lambda_{\star}/L)^{\beta+3/2}\, n^{\beta+3/2}

1927: \sim (\lambda_{\star}/\lambda)^{\beta+3/2},

1928: \eqno (22)

1929: $$

1930: where $\epsilon^{LW}_{\lambda}$

1931: is a relative amplitude of surface perturbation.

1932: It characterizes an angle

1933: of a deviation of a tangent to perturbed surface

1934: from unperturbed position.

1935: 

1936: 

1937: 

1938: 

1939: 

1940: At $\lambda\sim \lambda_{\star}$

1941: we have $ \eta^{LW}_{\lambda} \sim \lambda $ (22) -

1942: the assumption of linearity of perturbations is broken.

1943: As we are interested with long wavelengths

1944: $ \lambda> \lambda_{\star},$

1945: should be

1946: $$

1947: \beta>-3/2. \eqno (23)

1948: $$

1949: Value

1950: $$

1951: \beta = - 3/2 \eqno (24)

1952: $$

1953: corresponds to a {\it homogeneous} spectrum of perturbations

1954: $$

1955: \eta^{LW}_n\sim \epsilon_{hom} L/n^{3/2}\propto 1/|k_{\bot}|^{3/2},\;\;\;\;

1956: \eta^{LW}_{\lambda}\propto 1/n,\;\;\;\;

1957: \epsilon^{LW}_{\lambda}\sim \epsilon_{hom}={\rm const,} \eqno (25)

1958: $$

1959: where $k_{\bot}$ is transverse or horizontal wave number.

1960: Homogeneous noise (25)

1961: are dimensionless in the sense

1962: that do not contain parameter $ \lambda_{\star} $

1963: with dimension of length

1964: as in this case it is carried out

1965: $\eta^{LW}_{\lambda} \sim \lambda.$

1966: They are homogeneous in space of wave numbers

1967: as have no the allocated scale (from here the name).

1968: These noise

1969: play the important role in the theory of the turbulence

1970: connected to Rayleigh-Taylor instability.

1971: 

1972: 

1973: 

1974: 

1975: 

1976: 

1977: 

1978: 

1979: Let's "sew"

1980: a long wavelength part of the initial spectrum

1981: set by formulas (21), (22)

1982: with perturbations

1983: "located" at wave number axis

1984: $\lambda_{ini}, w_{\lambda ini}$\footnote{

1985: The short wavelength tail

1986: $ \lambda <\lambda_{ini}$

1987: is not significant for dynamics of system at $t\gg t_{\lambda ini}. $

1988: The long wavelength tail

1989: created by "located" at $k\sim k_{ini}$ perturbation (Fig. 15)

1990: looks like $w_n\propto n $ (15).

1991: 

1992: Long wavelength spectra (15) and (21) are different.

1993: }.

1994: Then amplitudes $w^{LW}_{\lambda}$

1995: (curves 1 and 3 in Fig. 19)

1996: and $w_{\lambda ini}$

1997: (a curve 4)

1998: are of the same order

1999: at $\lambda\sim\lambda_{ini}.$

2000: Let's write out a condition of sewing together

2001: $\epsilon^{LW}_{\lambda} \sim \eta_{\lambda ini}/\lambda_{ini}$

2002: having taken $\epsilon^{LW}_{\lambda}$ in this condition

2003: at $ \lambda\sim \lambda_{ini}.$

2004: Let's express parameter $ \lambda_{\star} $

2005: through $ \lambda_{ini} $ with the help of the given condition

2006: $$

2007: \lambda_{\star} \sim

2008: (\eta_{\lambda ini}/\lambda_{ini})^{1/(\beta+3/2)} \, \lambda_{ini}.

2009: $$

2010: We see,

2011: at $ \eta_{\lambda ini} <\lambda_{ini} $

2012: for the spectra limited to a condition (23)

2013: is carried out $ \lambda_{\star} <\lambda_{ini}, $ see Fig. 19.

2014: 

2015: 

2016: 

2017: 

2018: Excluding $ \lambda_{\star}$ it is received

2019: $$

2020: \eta^{LW}_n=\epsilon_{ini} \frac{

2021: \lambda_{ini}^{\beta+3/2}

2022: }

2023: {L^{\beta+1/2}}\,

2024: n^{\beta},        \;\;

2025: \eta^{LW}_{\lambda}=

2026: \epsilon_{ini} \frac{

2027: \lambda_{ini}^{\beta+3/2}

2028: }

2029: {L^{\beta+1/2}}\,

2030: n^{\beta+1/2},\;\;

2031: \epsilon^{LW}_{\lambda}=

2032: \epsilon_{ini} \left(\frac{

2033: \lambda_{ini}

2034: }

2035: {L}\right)^{\beta+3/2}\,

2036: n^{\beta+3/2}, \eqno (26)

2037: $$

2038: Where the notation

2039: $\epsilon_{ini} = \eta_{\lambda ini}/\lambda_{ini}$

2040: is entered.

2041: Equations (26)

2042: describe added\footnote{

2043: Added in the initial data

2044: to "located" on wave number $k$ initial perturbation.

2045: This is the combined

2046: ("located" and long wavelength)

2047: or compound perturbation.

2048: }

2049: long wavelength initial "noise".

2050: They replace expressions (21) and (22).

2051: 

2052: 

2053: 

2054: {\bf Development of Richtmyer-Meshkov turbulence

2055: at presence of added long wavelength "noise".}

2056: In Fig. 19 perturbation $\lambda_{ini}, w_{\lambda ini}$

2057: (a curve 4)

2058: "located" on an axis $n $

2059: is shown.

2060: Large-scale fluctuations of the "located" perturbation

2061: generate (\#\# 5-9) a long wavelength part of a spectrum.

2062: It is represented by a curve 2.

2063: Thus, the spectrum representing association of curves 2 and 4

2064: is connected to the "located" perturbation.

2065: 

2066: 

2067: 

2068: In Fig. 19

2069: added long wavelength "noise"

2070: is represented with curves 1 and 3

2071: concerning to different values of an index $\beta$ (26).

2072: For an example from a curve 1

2073: we have $\beta <0, $

2074: and for a curve 3 it is carried out $ \beta> 0.$

2075: At $ \beta=0 $

2076: dependences $ \epsilon_{\lambda} $ from $\lambda $

2077: are identical to added long wavelength "noise" (a curve 2)

2078: and for large-scale fluctuations of the "located" perturbation (a curve 2).

2079: Let's note,

2080: that under condition of (23)

2081: for these dependences the limiting equation

2082: $ \epsilon_{\lambda} \to 0$

2083: is carried out

2084: at $ \lambda\to\infty. $

2085: 

2086: 

2087: 

2088: Interest represents a case

2089: $-3/2\leq\beta <0$

2090: when added long wavelength "noise"

2091: surpasses large-scale fluctuations of the "located" perturbation.

2092: Such "strong" noise defines dynamics of mixing

2093: at $t\gg t_{\lambda ini}. $

2094: 

2095: 

2096: 

2097: 

2098: The harmonics of velocity

2099: corresponding to this "noise"

2100: are defined by Richtmyer formula (5)

2101: $$

2102: w^{LW}_n\sim (2\pi/L) n \eta^{LW}_n W_{unpertCB} \propto n^{1 +\beta}.

2103: $$

2104: For large-scale fluctuations of velocity it is received

2105: $$

2106: w_{\lambda} \sim \sqrt{n} w_n\propto n^{\beta+3/2}. \eqno (27)

2107: $$

2108: Let's note,

2109: that according to Richtmyer formula

2110: it is carried out $w_{\lambda} \propto\epsilon_{\lambda}. $

2111: 

2112: 

2113: 

2114: 

2115: Let's write down

2116: $dh_+/dt \sim w_{\lambda curr}$

2117: (see \# 8 and \# 9).

2118: We have

2119: $h_+\sim \lambda_{curr} $ (Figs. 6-8).

2120: From here we receive

2121: $$

2122: h_+\sim \lambda_{curr}\sim \epsilon_{ini}^{1/(\beta+5/2)}\,

2123: \lambda_{ini}^{(\beta+3/2)/(\beta+5/2)}\,

2124: W_{unpertCB}^{1/(\beta+5/2)}\,

2125: t^{1/(\beta+5/2)}

2126: \sim 

2127: $$$$

2128: \sim \lambda_{ini}^{(\beta+3/2)/(\beta+5/2)}\,

2129: w_{\lambda ini}^{1/(\beta+5/2)}\,

2130: t^{1/(\beta+5/2)},

2131: \eqno (28)

2132: $$

2133: as

2134: $w_{\lambda curr} \propto 1/\lambda_{curr}^{\beta+3/2}$

2135: according to (27).

2136: 

2137: 

2138: 

2139: 

2140: Homogeneous noise (24), (25) is

2141: the extreme case of the large-scale noise

2142: stimulating mixing.

2143: In this case

2144: during the initial moment

2145: we have $w_{\lambda} \sim w_{\lambda ini}$

2146: at anyone $ \lambda> \lambda_{ini}. $

2147: Hence, the formula (28) becomes $h_+\sim w_{\lambda ini} t. $

2148: 

2149: 

2150: 

2151: 

2152: {\bf Comparison of stimulated and spontaneous mixing.}

2153: There is a certain analogy

2154: between classes of turbulent movements

2155: at Rayleigh-Taylor and Richtmyer-Meshkov instabilities.

2156: There is a division on

2157: stimulated and spontaneous classes

2158: in Rayleigh-Taylor case

2159: (Inogamov, Demianov, and Son, 1999; Inogamov, 1999)

2160: and analogue of this division

2161: in Richtmyer-Meshkov case\footnote{

2162: According to it

2163: let's carry problem I (see the beginning \# 11)

2164: to spontaneous Richtmyer-Meshkov turbulence

2165: and a problem II - to stimulated.

2166: }.

2167: Let's remind sense of concepts

2168: about stimulated and spontaneous motions.

2169: At spontaneous turbulence

2170: unstable boundary

2171: is initially "seeded"

2172: with the perturbations

2173: "located" on $k $ or $ \lambda $ near $ \lambda_{ini}. $

2174: Are interested in an self-similar regime

2175: which is established asymptotically

2176: at $h_+\gg \lambda_{ini}. $

2177: Flow tends to a self-similar expansion regime

2178: of a mixing zone

2179: $$

2180: h_{\pm} = \alpha_{\pm} {\rm At} g t^2 \eqno (29)

2181: $$

2182: with factors $ \alpha_{\pm}^{sp}$

2183: not dependent on a concrete shape of "located" on $k$ initial perturbations

2184: ("forgeting" of initial perturbations $ \sim\lambda_{ini}).$

2185: 

2186: 

2187: 

2188: 

2189: We call stimulated a regime with long wavelength stimulation.

2190: Imposing at $t=0 $

2191: long wavelength power law spectra

2192: such as (26)

2193: introduces scale $\lambda_{\star}$

2194: and thus "breaks" self-similar dependence (29).

2195: The special place belongs to homogeneous "noise"

2196: with perturbations of boundary

2197: $ \eta $ a kind (25)

2198: and/or velocity perturbations of a kind

2199: $$

2200: w_n\sim \epsilon_{hom}^v \sqrt{{\rm At}gL}/n\propto 1/|k_{\bot}|,\;\;\;

2201: w_{\lambda}\sim \epsilon_{hom}^v \sqrt{{\rm At}g\lambda}. \eqno (30)

2202: $$

2203: In 3D a case

2204: (see also \# 12 below)

2205: we have the following expressions for homogeneous noise

2206: $$

2207: \eta_{nm}\sim \frac{\epsilon_{hom} L}{N_{nm}^2}\propto |k_{\bot}|^{-2},\;\;\;

2208: \eta_{\lambda}\sim \frac{\epsilon_{hom} L}{N_{nm}},\;\;\;

2209: w_{nm}\sim \epsilon_{hom}^v \frac{\sqrt{{\rm At}gL}}{N_{nm}^{3/2}}\propto

2210:  |k_{\bot}|^{-3/2}.

2211:  \eqno (8.30)'

2212: $$

2213: Homogeneous noise do not contain dimensional parameter

2214: $ \lambda_{\star}. $

2215: Hence, self-similarity (29) it is kept.

2216: However,

2217: with "noise" (25), (30), (30)'

2218: dimensionless numbers or amplitudes are connected

2219: $\epsilon_{hom}, \epsilon_{hom}^v. $

2220: In such situation

2221: factors $ \alpha_{\pm} $

2222: become functions

2223: $ \alpha_{\pm}^{st}(\epsilon_{hom}, \epsilon_{hom} ^v) $

2224: of these dimensionless amplitudes

2225: describing intensity of "noise" stimulating mixing.

2226: 

2227: 

2228: 

2229: 

2230: Real perturbation

2231: is a combination of spontaneous and stimulated\footnote{

2232: In cases with small values of stabilizing factors

2233: (viscosity, heat conductivity, diffusion)

2234: spontaneous turbulence gets ineradicable character

2235: while stimulated turbulence -

2236: it is supervised by dimensionless amplitudes $ \epsilon_{hom}. $

2237: }

2238: perturbations.

2239: Dependence

2240: $ \alpha_{\pm}^{st}(\epsilon_{hom}, \epsilon_{hom}^v) $

2241: is slowly growing function of the arguments

2242: (a curve "st" in Fig. 20).

2243: The factor

2244: $ \alpha_+^{sp+st}(\epsilon_{hom})$

2245: (a curve "sp+st")

2246: is limited from bottom\footnote{

2247: Due to reduction of $\epsilon_{hom}$

2248: it is impossible to achieve that

2249: mixing will develope more slowly than spontaneous mixing

2250: (the spontaneous regime "props up" the combined modes "from below").

2251: Similar situation exists

2252: for the Richtmyer-Meshkov mixing, see Fig. 21 below.}

2253: by the constant value

2254: $ \alpha_+^{sp} $ (a straight line "sp").

2255: From the data on factor $ \alpha_+^{sp}$

2256: suggested in report Youngs 2004

2257: in Fig. 20 value of $0.03$ is accepted.

2258: There is a value of dimensionless or relative amplitude

2259: $ \epsilon_{hom}^{thr}$

2260: at which influence of stimulation becomes appreciable.

2261: 

2262: 

2263: 

2264: 

2265: 

2266: It is very important

2267: that numerically the value

2268: $ \epsilon_{hom}^{thr} $ is rather small.

2269: It was supposed in the previous estimations

2270: (Inogamov, Demianov, and Son, 1999; Inogamov, 1999)

2271: and has been confirmed in recent detailed numerical calculations

2272: (Youngs 2004).

2273: Smallness of the factor $\alpha_+^{sp}$

2274: and smallness of the threshold amplitude

2275: $\epsilon_{hom}^{thr}$

2276: strongly raise a role of a regime with stimulation.

2277: The amplitude $ \epsilon_{hom} $

2278: gives value of deviations

2279: of a tangent to boundary on all scales

2280: as at homogeneous noise $ \epsilon_{\lambda} $

2281: does not depend from $ \lambda. $

2282: Smallness of $\epsilon_{hom}^{thr}$ means

2283: that hidden or latent widezone perturbations are significant.

2284: It is caused by that

2285: these weak widezone noise can increase

2286: essentially

2287: the major characteristic of mixing - the factor $ \alpha_+. $

2288: 

2289: 

2290: 

2291: In reports Youngs 2004 and Dimonte et al. 2004

2292: it is assumed,

2293: that available divergences

2294: between numerical $ (\alpha_+\approx 0.03) $

2295: and experimental $ (\alpha_+\approx 0.05) $

2296: estimations $ \alpha_+ $

2297: are caused by the latent stimulation of mixing in experiment.

2298: Such stimulation difficultly to supervise

2299: because of smallness of $ \epsilon_{hom}^{thr}. $

2300: The origin of $ \epsilon_{hom} $

2301: can be connected to vibration of experimental installations.

2302: The interesting program

2303: (Dimonte et al., 2004)

2304: for search of weak widezone noise in experiment is offered.

2305: 

2306: 

2307: 

2308: 

2309: 

2310: {\bf Analogue of stimulation of mixing in Richtmyer-Meshkov case.}

2311: At Richtmyer-Meshkov turbulence

2312: large-scale perturbations (26) intensify mixing.

2313: In stimulated regime

2314: (fig. 21)

2315: the exponent $ \theta $ is more,

2316: than in a spontaneous regime:

2317: $ \theta^{st}> \theta^{sp}, $

2318: $ \theta^{st} = (\beta+5/2)^{-1}, $

2319: see (28) \footnote{

2320: In report Youngs 2004

2321: stimulation of Richtmyer-Meshkov mixing is numerically investigated.

2322: Power law initial surface perturbations of a kind (26) has been used.

2323: Change of the exponent $\theta$

2324: is revealed at change of an index $\beta. $

2325: },

2326: $\theta^{sp} =2/5, $ see (17), \# 8, \# 9.

2327: Borders of an interval of values of an index $ \beta $

2328: form points $ \beta=0 $ and $ \beta = - 3/2 $ on an axis $ \beta. $

2329: In Fig. 21 more to the right of the specified interval

2330: the forbidden area, see (23) settles down.

2331: More to the left - the area lays,

2332: in which stimulation it is possible to neglect

2333: (compare spectra 2 and 3 in Fig. 19).

2334: The point $ \beta = - 3/2 $ (24)

2335: represents homogeneous or "dimensionless" "noise" (25).

2336: To them there corresponds

2337: the maximal value of function $ \theta^{st}(\beta.) $

2338: 

2339: 

2340: 

2341: 

2342: 

2343: 

2344: The most important\footnote{

2345: As this point represents the most widespread variant of initial conditions

2346: With "localization" of perturbations

2347: on wave number $k $ (a random chain $ \lambda_{ini}). $

2348: }

2349: is the point $ \beta=0$.

2350: It

2351: concerns to a mode

2352: which we shall conditionally name "spontaneous".

2353: The reserve here contains in the following.

2354: The comparative analysis

2355: of Rayleigh-Taylor and Richtmyer-Meshkov problems

2356: is carried out.

2357: At Rayleigh-Taylor turbulence

2358: in a spontaneous case

2359: there is no dynamically essential long-wave tail

2360: existing on "all times" $t> 0 $

2361: (to it is connected "forgeting"

2362: small-scale initial perturbations).

2363: On the contrary,

2364: in the stimulated case

2365: such tail is present

2366: (constant presence of a part $ \lambda> \lambda _ {curr} $

2367: of an initial spectrum).

2368: In this respect

2369: at Richtmyer-Meshkov mixing business is in another way.

2370: Namely,

2371: both in stimulated and in spontaneous cases

2372: there is constantly existing long wavelength tail

2373: determining dynamics.

2374: Hence, and in a spontaneous case

2375: the certain "trace" of the initial data is kept.

2376: 

2377: 

2378: 

2379: Richtmyer-Meshkov turbulence

2380: in a regime (17)

2381: is generated by random "located" on $k$ perturbation

2382: (\#\# 5-8).

2383: Such perturbation

2384: generates spontaneous mixing in Rayleigh-Taylor case.

2385: The basis is connected to it

2386: to name spontaneous a regime of mixing (17) in Richtmyer-Meshkov case.

2387: 

2388: 

2389: 

2390: 

2391: \vspace{.5cm}

2392: \centerline{{\bf \# 12. 3D case}}

2393: 

2394: 

2395: 

2396: \vspace{.5cm}

2397: Let's generalize the description of turbulent mixing

2398: received above on 3D space.

2399: The horizontal plane with coordinates $x, y $

2400: (Fig. 2) is transverse plane.

2401: Let's limit a part of this plane

2402: by big

2403: $ (L\gg\lambda_{ini}) $

2404: square $L\times L $ ("box", \# 2).

2405: Let's consider

2406: a random two-dimensional lattice of perturbations

2407: on a plane $x, y. $

2408: Such lattice (3D)

2409: replaces a random chain (2D, \# 5).

2410: Let the average transversal size

2411: of initial surface perturbations

2412: $$

2413: z = \eta_{ini}(x, y) \eqno (31)

2414: $$

2415: in this lattice

2416: is equal to $\lambda_{ini}$

2417: and randomness is characterized by good quality $Q\sim 1$

2418: (\# 5).

2419: 

2420: 

2421: 

2422: 

2423: 

2424: Let's expand

2425: periodic in a square $L\times L $ function (31)

2426: in Fourier series

2427: $$

2428: \eta_{ini}(x,y)=\sum_{n=1}^{\infty}\sum_{m=1}^{\infty} (\eta_{nm}C_{nx}C_{my}+

2429: \eta_{nm}'C_{nx}S_{my}+\eta_{nm}''S_{nx}C_{my}+\eta_{nm}'''S_{nx}S_{my}),

2430:  \eqno (32)

2431: $$

2432: Where $C_{nx} \equiv \cos (2\pi n x/L), $

2433: $S_{nx} \equiv \sin (2\pi n x/L). $

2434: Amplitudes of Fourier harmonics

2435: in series (32) are calculated by integration

2436: $$

2437: \eta_{nm}=(4/L^2)\int_0^L\int_0^L dx \,dy\,\eta_{ini}(x,y)C_{nx}C_{my}. \eqno (33)

2438: $$

2439: Similar formulas

2440: define other amplitudes which are included in a series (32).

2441: 

2442: 

2443: 

2444: 

2445: 

2446: 

2447: 

2448: 

2449: 

2450: Long wavelength $ (n, m\ll N_{ini} =L/\lambda_{ini})$

2451: asymptotics of spectra $ \eta_{nm} $ looks like

2452: $$

2453: \eta_{nm}\sim \xi_{nm} \cos\psi_{nm}

2454: (4/L^2) \eta_{rms} \lambda_{ini}^2

2455: \sqrt{\lambda^2/\lambda_{ini}^2}

2456: \sqrt{L^2/\lambda^2}\sim \eta_{rms}\lambda_{ini}/L\sim \eta_{rms}/N_{ini},

2457: \eqno (34)

2458: $$

2459: where $ \langle\xi_{nm} \rangle=1, $ $ \psi_{nm} $ is a random phase,

2460: $ \eta_{rms}^2 = (1/L^2) \int_0^L\int_0^L dx \, dy \,\eta_{ini}^2, $

2461: $ \lambda=2\pi/k, $ $k^2 = (k^x_n)^2 + (k^y_m)^2, $

2462: $k^x_n=2\pi n/L, $ $k^y_m=2\pi m/L $ are components of a wave vector

2463: on a plane $ (x, y). $

2464: This vector corresponds to a harmonic $ (n, m). $

2465: Asymptotics (34)

2466: follows from a statistical estimation of integral (33)

2467: of stochastic function (31).

2468: Occurrence of statistical factors

2469: is connected to stochastic fluctuations of this function

2470: $\sqrt{\lambda^2/\lambda_{ini}^2} $ and $ \sqrt{L^2/\lambda^2}. $

2471: The estimation (34) is valid on an interval of values $n, m $

2472: from 1 up to $ \sim N_{ini}. $

2473: 

2474: 

2475: 

2476: In 3D a case instead of (7) we have

2477: $\eta_{\lambda}\sim \sqrt{n^2+m^2}\eta_{nm}\sim \eta_{rms}\lambda_{ini}/\lambda,$

2478: where $n\sim m\sim L/\lambda. $

2479: 

2480: 

2481: 

2482: 

2483: Let perturbation (31) is linear (small amplitude):

2484: $ \eta_{rms} \ll \lambda_{ini}. $

2485: Let's estimate velocity field after passage of a shock wave.

2486: We use Richtmyer formula

2487: connecting amplitudes of harmonics of surface perturbations

2488: $ \eta_{nm} $ and vertical velocity $w_{nm}. $

2489: In 3D geometry we have

2490: $$

2491: w_{nm}=F {\rm At}^+ (2\pi/L) \sqrt{n^2+m^2} \eta^+_{nm} W_{unpertCB}, \eqno (8.35)

2492: $$

2493: compare with the formula (5).

2494: 

2495: 

2496: 

2497: 

2498: Let's find velocity of movement of primary bubbles $w_{\lambda ini}. $

2499: Such bubbles have transverse scale $ \sim\lambda_{ini}. $

2500: Let's substitute in the formula (35)

2501: expression for amplitude $ \eta_{nm} $ (34) at $n\sim m\sim N_{ini}. $

2502: In result it is received

2503: $$

2504: w_{nm}(n,m\sim N_{ini})\sim F {\rm At}^+ (\eta_{rms}/L) W_{unpertCB}. \eqno (36)

2505: $$

2506: 

2507: 

2508: 

2509: 

2510: Let's write down

2511: $$

2512: w_{rms}\sim w_{\lambda ini}\sim

2513:  w_{\lambda}(\lambda\sim\lambda_{ini})\sim N_{ini} w_{nm}(n,m\sim N_{ini}) \eqno (37)

2514: $$

2515: and

2516: $$

2517: w_{\lambda}(\lambda\sim\lambda_{nm}\gg\lambda_{ini})\sim (L/\lambda_{nm}) w_{nm},\;\;\;\;

2518: \lambda_{nm}=L/\sqrt{n^2+m^2}. \eqno (38)

2519: $$

2520: In the formula (38)

2521: size $w_{\lambda} $

2522: is determined by the formula (6) generalized on 3D a case.

2523: Formulas (37), (38) replace the formula (7) in 3D a case.

2524: Fourier amplitudes $w_n $ (2D) or $w_{nm} $ (3D)

2525: are connected to the corresponding size $w_{\lambda} $

2526: by means of the statistical factor

2527: of order of a root square from number of modes of scale $ \lambda. $

2528: In 2D case this number is $L/\lambda$

2529: and in 3D a case - is $ (L/\lambda)^2. $

2530: 

2531: 

2532: 

2533: 

2534: 

2535: From formulas (36), (37)

2536: it is possible to define velocity of primary bubbles $w_{\lambda ini}. $

2537: The estimation of this velocity looks like

2538: $$

2539: w_{\lambda ini}\sim F {\rm At}^+ N_{ini} (\eta_{rms}/L) W_{unpertCB}. \eqno (39)

2540: $$

2541: 

2542: 

2543: 

2544: 

2545: Asymptotics of 3D Richtmyer-Meshkov mixing

2546: at late times is defined by a long wavelength part

2547: $ (n, m\ll N_{ini}) $ of a spectrum of velocity $w_{nm}. $

2548: Let's find asymptotics of a spectrum $w_{nm} $

2549: with the help of asymptotics of a spectrum $ \eta_{nm} $ (34),

2550: Richtmyer formula (35) and auxiliary expression (39).

2551: We have

2552: $$

2553: w_{nm}\sim (\sqrt{n^2+m^2}/N_{ini}^2) w_{\lambda ini}. \eqno (40)

2554: $$

2555: It is useful to compare 3D expression (40) to the formula (15)

2556: valid in 2D case.

2557: 

2558: 

2559: 

2560: From formulas (38) and (40)

2561: the asymptotics $(\lambda\gg \lambda_{ini}) $ of velocity fluctuations

2562: of scale $ \lambda $

2563: can be found.

2564: It may be written as

2565: $$

2566: w_{\lambda}(\lambda\sim\lambda_{nm}\gg\lambda_{ini})\sim \frac{n^2+m^2}{N_{ini}^2} w_{\lambda ini}\sim

2567: \left(\frac{\lambda_{ini}}{\lambda}\right)^2 w_{\lambda ini}. \eqno (41)

2568: $$

2569: 

2570: 

2571: 

2572: 

2573: The hierarchy (13) of fluctuations (41)

2574: of different scales

2575: forms enlarging cell structure

2576: of velocity field $w(x,y,z)$

2577: similar to presented in Fig. 17.

2578: Three-dimensional cells

2579: have the sizes $ \sim\lambda\times\lambda\times\lambda. $

2580: As well as in 2D a case,

2581: decrease of amplitude of fluctuations of 3D fields $w(x, y, z) $

2582: at removal from boundary

2583: is described by the power law.

2584: To tell the truth, an exponent is another.

2585: From the formula (41) it is received

2586: $$

2587: w(z)\sim w_{\lambda ini} (\lambda_{ini}/z)^2, \eqno (42)

2588: $$

2589: compare with (14).

2590: 

2591: 

2592: 

2593: 

2594: Decay of velocity $w(z) $

2595: defines the law of penetration of mixing front $h_+(t)$.

2596: Let's write down $dh_+/dt \sim w(z\sim h_+)$ (see \# 8).

2597: Substituting here the formula (42) and integrating

2598: we receive

2599: $$

2600: h_+\sim w_{\lambda ini}^{1/3} \lambda_{ini}^{2/3} t^{\theta},

2601: \;\;\;\;\theta=1/3. \eqno (43)

2602: $$

2603: Let's compare laws (43) and (17)

2604: concerning to 2D and 3D to cases.

2605: We see, in a three-dimensional case

2606: Richtmyer-Meshkov turbulence fades faster

2607: and the front $h_+(t) $ moves ahead essentially more slowly

2608: $$

2609: \theta _ {3D} <\theta _ {2D},

2610: $$

2611: where values $ \theta_{2D} $ and $ \theta_{3D} $

2612: are given by formulas (17) and (43) accordingly.

2613: 

2614: 

2615: 

2616: 

2617: \newpage

2618: \centerline{{\bf References}}

2619: 

2620: 

2621: \noindent M. Chertkov, 2003,

2622: Phenomenology of Rayleigh-Taylor turbulence,

2623: Phys. Rev. Lett., vol. 91(11), 115001 (2003).

2624: 

2625: 

2626: 

2627: \noindent V. Demchenko, 2004,

2628: Colliding surface instability for a high velocity impact,

2629: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2630: Cambridge, UK, 19-23 July 2004, Abstracts, p. 24;

2631: http://www.damtp.cam.ac.uk/iwpctm9/

2632: 

2633: 

2634: \noindent G. Dimonte, P. Ramaprabhu, and M. Andrews, 2004,

2635: Dependence of self-similar Rayleigh-Taylor growth on initial conditions,

2636: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2637: Cambridge, UK, 19-23 July 2004, Abstracts, p. 26;

2638: http://www.damtp.cam.ac.uk/iwpctm9/

2639: 

2640: 

2641: 

2642: \noindent N.A. Inogamov, 1999,

2643: The Role of Rayleigh-Taylor and Richtmyer-Meshkiv Instabilities in Astrophysics: An Introduction,

2644: Astrophysics and Space Physics Reviews, vol. 10, part 2, 1-335 (1999).

2645: 

2646: 

2647: \noindent N.A. Inogamov, A.Yu. Demianov, and E.E. Son, 1999,

2648: Hydrodynamics of mixing:

2649: Periodical structures,

2650: Amplification of subharmonics

2651: and inverse cascad,

2652: in Russian,

2653: Moscow, MIPT, 1999, 464 p.

2654: 

2655: 

2656: 

2657: \noindent N.A. Inogamov and A.M. Oparin, 1999,

2658: Three-dimensional array structures associated with

2659: Richtmyer–Meshkov and Rayleigh–Taylor instability,

2660: Journal of Experimental and Theoretical Physics, vol. 89(3), 481-499 (1999).

2661: 

2662: 

2663: 

2664: \noindent N.A. Inogamov, M. Tricottet, A.M. Oparin, and S. Bouquet, 2001a,

2665: Three-Dimensional Morphology of Vortex Interfaces

2666: Driven by Rayleigh-Taylor or Richtmyer-Meshkov Instability,

2667: arXiv:physics/0104084 (2001).

2668: 

2669: 

2670: 

2671: \noindent N.A. Inogamov, A.M. Oparin, A.Yu. Dem'yanov, L.N. Dembitskii, and V.A. Khokhlov, 2001b,

2672: On stochastic mixing caused by the Rayleigh-Taylor instability,

2673: Journal of Experimental and Theoretical Physics, vol. 92(4), 715-743 (2001).

2674: 

2675: 

2676: 

2677: \noindent N.A. Inogamov, 2002,

2678: Statistics of long wavelength fluctuations and expansion law

2679: for turbulent mixing zone driven by Richtmyer-Meshkov instability,

2680: Pis'mav ZhETF (JETP Letters), vol. 75(11), 664-668 (2002).

2681: 

2682: 

2683: 

2684: 

2685: \noindent Yu.A. Kucherenko, S.I. Balabin, R.I. Ardashova, O.E. Kozelkov, A.V. Dulov, and I.A. Romanov,

2686: 2004,

2687: On possibility of experimental determination of integral characteristics of molecular component of mixing

2688: by means of X-ray technique,

2689: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2690: Cambridge, UK, 19-23 July 2004, Abstracts, p. 57;

2691: http://www.damtp.cam.ac.uk/iwpctm9/

2692: 

2693: 

2694: 

2695: \noindent Yu.A. Kucherenko, A.P. Pylaev, V.D. Murzakov, A.V. Belomestnih, V.N. Popov, and A.A. Tyaktev,

2696: 2004,

2697: On possibility of experimental determination of molecular component of non-stationary turbulent mixing,

2698: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2699: Cambridge, UK, 19-23 July 2004, Abstracts, p. 58;

2700: http://www.damtp.cam.ac.uk/iwpctm9/

2701: 

2702: 

2703: 

2704: \noindent P.F. Linden, J.M. Redondo, and D. Youngs, 1994,

2705: Molecular mixing in Rayleigh-Taylor instability,

2706: Journal of Fluid Mechanics, vol. 265, 97-124 (1994).

2707: 

2708: 

2709: 

2710: 

2711: \noindent J.M. Redondo and G. Garzon, 2004,

2712: Multifractal structure and intermittent mixing in Rayleigh-Taylor driven fronts,

2713: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2714: Cambridge, UK, 19-23 July 2004, Abstracts, p. 91;

2715: http://www.damtp.cam.ac.uk/iwpctm9/

2716: 

2717: 

2718: 

2719: 

2720: \noindent O. Schilling, M. Latini, W.-S. Don, and B. Bihari, 2004,

2721: Investigation of the large-scale and statistical properties

2722: of Richtmyer-Meshkov instability-induced mixing,

2723: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2724: Cambridge, UK, 19-23 July 2004, Abstracts, p. 97;

2725: http://www.damtp.cam.ac.uk/iwpctm9/

2726: 

2727: 

2728: 

2729: 

2730: \noindent D. Shvarts, D. Oron, D. Kartoon, et. al., 2000,

2731: Scaling laws of nonlinear Rayleigh-Taylor and Richtmyer-Meshkov instabilities

2732: in two and three dimensions,

2733: C.R. Acad. Sci. Paris, t. 1, Serie IV, 719-726 (2000).

2734: 

2735: 

2736: 

2737: 

2738: 

2739: \noindent J.G. Wouchuk, 2001a,

2740: Growth rate of the linear Richtmyer-Meshkov instability

2741: when a shock is reflected,

2742: Phys. Rev. E, vol. 63, 056303 (2001a).

2743: 

2744: 

2745: 

2746: \noindent J.G. Wouchuk, 2001b,

2747: Growth rate of the Richtmyer-Meshkov instability

2748: when a rarefaction is reflected,

2749: Phys. of Plasmas, vol. 8(6), 2890-2907 (2001b).

2750: 

2751: 

2752: 

2753: \noindent D.L. Youngs, 2004,

2754: Effect of initial conditions on self-similar turbulent mixing,

2755: 9th International Workshop on the Physics of Compressible Turbulent Mixing,

2756: Cambridge, UK, 19-23 July 2004, Abstracts, p. 123;

2757: http://www.damtp.cam.ac.uk/iwpctm9/

2758: 

2759: 

2760: 

2761: \newpage

2762: \centerline{{\bf Figures}}

2763: 

2764: \begin{center}

2765: \includegraphics[width=12cm]{Fig1.ps}

2766: 

2767: Fig. 1. Unperturbed flow.

2768: It is shown

2769: approach of a shock wave (a straight line 2 and an arrow)

2770: to contact boundary (a straight line 1).

2771: \end{center}

2772: 

2773: \begin{center}

2774: \includegraphics[width=12cm]{Fig2.ps}

2775: 

2776: Fig. 2. Richtmyer-Meshkov instability

2777: arises from surface perturbations

2778: (a continuous curve).

2779: Unpertubed position of boundary is shown by dashed straight line.

2780: Definition of longitudinal and transverse coordinates.

2781: \end{center}

2782: 

2783: \begin{center}

2784: \includegraphics[width=12cm]{Fig3.ps}

2785: 

2786: \end{center}

2787: 

2788: Fig. 3. Baroclinic "processing" by a shock wave

2789: of surface perturbations $ \eta $

2790: into velocity field $ \vec v. $

2791: The form (continuous curves)

2792: and directions of movement (arrows) of contact boundary

2793: after passage of a shock for cases in a phase (at the left)

2794: and in a counterphase (on the right).

2795: Shaped straight line is position of unperturbed boundary.

2796: \newpage

2797: 

2798: \begin{center}

2799: \includegraphics[width=12cm]{Fig4.ps}

2800: 

2801: 

2802: Fig. 4. The most interesting example

2803: (by virtue of the typicalness)

2804: of the general stochastic function

2805: set in scales $ \lambda_{ini} $ and $w_{\lambda ini} $

2806: $(w $ is a component of velocity on an axis $z). $

2807: The sign-variable field of velocity

2808: in the near boundary layer

2809: arising after passage of a shock

2810: through the perturbed boundary is presented.

2811: \end{center}

2812: 

2813: \begin{center}

2814: \includegraphics[width=12cm]{Fig5.ps}

2815: 

2816: 

2817: Fig. 5. The average distribution of light "l" and heavy "h" phases

2818: (a) prior to the beginning of mixing

2819: and (c) after the beginning of mixing.

2820: As a result of interpenetration of phases

2821: intermediate layer "M" with a mixture of "l" and "h" phases is formed.

2822: \end{center}

2823: 

2824: \begin{center}

2825: \includegraphics[width=11cm]{Fig6.ps}

2826: 

2827: 

2828: Fig. 6. Involving into turbulent movement of volumes of light "l"

2829: and heavy "h" liquids.

2830: \end{center}

2831: 

2832: \begin{center}

2833: \includegraphics[width=12cm]{Fig7.ps}

2834: 

2835: Fig. 7. (a) The instant picture alternating "l" and "h" columns.

2836: (b) Sign variability of velocity $w(x, y_{fix}, z=0, t_{fix}). $

2837: The plane $z=0 $ corresponds to position of unperturbed contact.

2838: \end{center}

2839: 

2840: \begin{center}

2841: \includegraphics[width=12cm]{Fig8.ps}

2842: 

2843: 

2844: Fig. 8. Large-scale structure and a vortical chain.

2845: (a) "Blowing" of "l" and "h" columns by pairs of vortical spots.

2846: (b) A chain of large-scale sign-variable vorticities

2847: or a chain of dipoles $ \lambda\times h $ from pairs of vorticities.

2848: \end{center}

2849: 

2850: \begin{center}

2851: \includegraphics[width=12cm]{Fig9.ps}

2852: 

2853: 

2854: Fig. 9. Formation of a claster dipole

2855: from small dipoles of the previous stages.

2856: \end{center}

2857: 

2858: \begin{center}

2859: \includegraphics[width=12cm]{Fig10.ps}

2860: 

2861: Fig. 10. Compound character of large-scale dipoles at late stages of mixing.

2862: \end{center}

2863: 

2864: \begin{center}

2865: \includegraphics[width=12cm]{Fig11.ps}

2866: 

2867: Fig. 11. Extension and deformation of dipoles of generation $i_1 $

2868: inside dipoles of the subsequent generation $i_2. $

2869: \end{center}

2870: 

2871: \begin{center}

2872: \includegraphics[width=12cm]{Fig12.ps}

2873: 

2874: 

2875: Fig. 12. Squeezing $ (p_{\infty}> 0) $ of bulk $ (+ \infty) $ of liquids

2876: by ram pressure of bubbles pressed into it.

2877: (a) $p(z) $ on a straight line (b$\infty); $

2878: (b) Increase of $p $ on isobars 0 (0 is a boundary of a bubble "bj"), 1, 2 and 3.

2879: At boundary "bj" $p_{CB} =0. $

2880: (c) $p(z) $ on (j$\infty). $

2881: 

2882: \end{center}

2883: 

2884: \begin{center}

2885: \includegraphics[width=12cm]{Fig13.ps}

2886: 

2887: 

2888: Fig. 13. (a) The layer with the turbulent motion

2889: concentrated near a plane $z=0. $

2890: (b) A profile $p(z) $ with "hole" $ (\mu\neq 0) $ in the region of movement.

2891: \end{center}

2892: 

2893: \begin{center}

2894: \includegraphics[width=11.5cm]{Fig14.ps}

2895: 

2896: 

2897: Fig. 14. (a) Penetration of bubble. Geometry of flow.

2898: (b) Distribution $p $ lengthways (b$\infty). $

2899: (c) An instant field of pressure.

2900: \end{center}

2901: 

2902: \begin{center}

2903: \includegraphics[width=12cm]{Fig15.ps}

2904: 

2905: 

2906: Fig. 15. Ensemble of bubbles $\lambda_i $

2907: with the average size $ \lambda_{ini} $

2908: and groups of bubbles, got on pieces $ \lambda $

2909: ($\lambda$-groups).

2910: \end{center}

2911: 

2912: \begin{center}

2913: \includegraphics[width=12cm]{Fig16.ps}

2914: 

2915: 

2916: 

2917: Fig. 16. (a) Changeability of pressure (pieces 3),

2918: created by $"\lambda$-groups" of bubbles.

2919: (b) Random alternation of squares $ \lambda\times\lambda $

2920: with positive and negative (are shaded) increments of velocity

2921: $w_{\lambda}. $

2922: \end{center}

2923: 

2924: \begin{center}

2925: \includegraphics[width=12cm]{Fig17.ps}

2926: 

2927: 

2928: Fig. 17. Hierarchy of changes of a sign

2929: of velocity $w (x, z): $

2930: small-scale changeability near surface,

2931: large-scale - it is far in a bulk of liquid.

2932: \end{center}

2933: 

2934: \begin{center}

2935: \includegraphics[width=12cm]{Fig18.ps}

2936: 

2937: 

2938: Fig. 18. (a) Scales $n-1, n, n+1 $ and the squares connected to them.

2939: (b) "Designing" the centers of squares from the left figure on right.

2940: \end{center}

2941: 

2942: \begin{center}

2943: \includegraphics[width=12cm]{Fig19.ps}

2944: 

2945: 

2946: Fig. 19. The combined perturbations

2947: from (1) added long wavelength "noise" (a curve 1)

2948: and (2) "located" on $ \lambda $ a spectrum (a curve 4).

2949: Perturbations 4 generate a large-scale tail (a curve 2).

2950: \end{center}

2951: 

2952: \newpage

2953: \begin{center}

2954: \includegraphics[width=12cm]{Fig20.ps}

2955: 

2956: 

2957: Fig. 20. Dependence of factor $\alpha_+^{sp+st}$

2958: for the combined perturbations

2959: from relative amplitude of long wavelength homogeneous noise

2960: $ \epsilon_{hom}. $

2961: \end{center}

2962: 

2963: \begin{center}

2964: \includegraphics[width=12cm]{Fig21.ps}

2965: 

2966: 

2967: Fig. 21. Analogue of Fig. 20 in Richtmyer-Meshkov case:

2968: $ \theta^{sp} $ limits an interval of $ \theta $ from below.

2969: \end{center}

2970: 

2971: 

2972: 

2973: 

2974: 

2975: 

2976: 

2977: 

2978: 

2979: \end{document}

2980: