1: \section{Boundary layers} \label{sec:bl}
2:
3: \subsection{Relevance of boundary layers and challenges}
4:
5: Boundary layers describe the temperature and flow fields in the vicinity of the plates and walls.
6: They are
7: characterized by their time-averaged profiles in the direction perpendicular to the respective solid boundary, i.e.,
8: in the $z$-direction off the bottom and top plates or in the $x$-direction off the side walls.
9: As the top and bottom
10: boundary layers contribute the main resistance for the heat transfer
11: through the cell and thus dominantly determine the Nusselt number,
12: they deserve special attention. Indeed, nearly all theories
13: of Nu(Ra,Pr) in RB convection are in essence boundary-layer theories:
14: This holds for the now classical mixing-layer theory of \cite{cas89},
15: the turbulent-BL based scaling theory of \cite{shr90} and extensions
16: thereof (e.g.\ \cite{chi97}), the turbulent BL type
17: theories of \cite{dub01,dub02b}
18: and of \cite{hoe05,hoe06}, and the theory of \cite{gro00,gro01,gro02,gro04} which
19: scaling-wise builds heavily on the laminar, though time dependent
20: Prandtl-Blasius BL theory as elaborated in subsection \ref{nuregl}.
21: It is therefore of prime importance to {\it directly} study the
22: BLs, to see whether the assumptions on which these theories are based are
23: fulfilled at least to a reasonable approximation.
24:
25: Unfortunately, directly studying the BLs in the large-Ra regime
26: is equally challenging
27: experimentally, numerically, and theoretically. Experimentally,
28: the required spatial resolution is very difficult to achieve. Even
29: for the $L \approx 6.5$ m high Ilmenau RB barrel
30: both the thermal and the kinetic BL thicknesses
31: are only a few millimeters when e.g. $\Ra \approx 10^{12}$, see \cite{pui07b,pui07}.
32: In addition, both laser-Doppler velocimetry (LDV) and hot-wire anemometry
33: are notoriously difficult to employ in regions with large temperature
34: fluctuations and a small mean-flow velocity; both complications occur in the BLs.
35: Numerically, large Rayleigh numbers are hard to reach,
36: in particular with sufficient spatial resolution, see section \ref{nuredns}. Even state-of-the-art
37: DNS such as those of \cite{ver08} have only a few gridpoints in the BLs. Theoretically,
38: no generally accepted BL theory for flow over a strongly heated surface
39: exists. Even in the truely laminar case an analytic theory for the two-way
40: coupling
41: of the
42: temperature to the velocity field is missing
43: (i.e., taking the temperature as an {\it active} scalar),
44: see e.g.\ \cite{sch00}. And even if the two-way coupling is suppressed
45: as done in the interesting numerical work by
46: \cite{chi01a} and \cite{chi02}, no exact analytical
47: results exist for the thermal BL thickness
48: and thus for the heat transfer through the shear flow over the heated plate.
49:
50: The analysis of the BLs is further complicated by their extreme
51: complexity, not only in time, where there is plume detachment (see figure \ref{plumes}),
52: but also in space. Recent experimental studies such as those by
53: \cite{lui98,wan03,may06}, and \cite{sun08} and
54: numerical studies such as those by \cite{sug08} give increasing evidence that
55: the BL thicknesses as well as the
56: profiles of the mean quantities and of the fluctuations
57: depend on the position along and above the plate(s),
58: not only relative to the walls, but also relative to the fluctuating
59: large convection roll (see section \ref{nurere}). It is this main convection roll
60: which creates the viscous BL because of the no-slip condition at the top and
61: bottom plates.
62:
63: The aim of this section is to give an overview of what has been found experimentally,
64: theoretically, and numerically about the BL thicknesses.
65:
66:
67:
68: \subsection{Thermal boundary layers}
69: The thermal boundary layer thickness can be defined in several ways.
70: From an experimental point of view, it is easiest to
71: time-average the temperature profile
72: at fixed lateral positions $(x,y)=*$ for various $z$ and to extract
73: a thermal boundary layer thickness $\lambda_\theta (*)$
74: from the resulting local profile.
75: From the
76: theoretical viewpoint
77: area-averages on $x,y$ (in addition to the time-averaging)
78: are of main interest. Indeed, it is the thickness
79: $\lambda_{\theta}$ of the
80: {\em area-averaged} temperature profile in the $z$-direction which enters theory
81: while comparison with experiment so far refers to locally
82: measured BL thicknesses $\lambda_{\theta}(*)$.
83: Note that strictly speaking the global thickness $\lambda_{\theta}$ is
84: not equal to
85: $\langle \lambda_{\theta}(*) \rangle_A$ in general; the reason is that the thickness is a (nonlinearly defined)
86: property of a given profile. Then
87: $\lambda_{\theta}$ in particular is a property of the area-averaged
88: profile and not the average of properties $\lambda_{\theta}(*)$ of the respective local profiles. Along large plates
89: the local BL width may well represent the area-averaged global one; but for RB samples with $\Gamma$ of order 1
90: one has to expect significant differences, even different scaling of the global and the local thicknesses,
91: since the area-averaged profiles comprise all
92: inhomogeneities of the temperature and flow fields along the plates, including backflows, the regions between them,
93: and the near wall ranges. These all will locally have quite different profiles. The more homogeneous the flow,
94: the more similiar
95: the global width will be to the average of the local widths. Experimental analysis of all these
96: details is still a big challenge.
97:
98: The (global) thermal BL thickness $\lambda_\theta$ can be defined in several ways.
99: The most popular way is to define $\lambda_\theta$ through the
100: slope of the area-averaged time-mean temperature profile at the plates:
101: Take the tangent of
102: the area-averaged mean temperature $\theta(z)$ at the plate.
103: That distance between the plate and the vertical position
104: where this tangent crosses the bulk (or center) temperature $T_c$ is then called
105: $\lambda_\theta^{sl}$. The center temperature $T_c$ is equal to the mean temperature
106: $T_m = (T_t + T_b)/2$ if the Oberbeck-Boussinesq approximation holds because of up-down-symmetry.
107: The very notion of $T_c$ already refers to an area-averaged profile.
108:
109: Locally, i.e., for fixed $(x,y) \equiv *$, the slope of the time-averaged profile, as well as the temperature
110: for $z \rightarrow \infty$ and thus the local width $\lambda_{\theta}^{sl}(*)$, will depend on the
111: horizontal position $*$ where the local time-mean temperature profile is taken.
112:
113: One of the first measurements of temperature and also velocity
114: profiles in RB cells was done by \cite{til93}, namely in water ($\Pr = 6.6$)
115: at fixed $\Ra = 1.1 \cdot 10^9$ and at a fixed lateral position $*$.
116: % over the plate.
117: \cite{bel93} extended these measurements to the $\Ra$-range
118: $5\cdot 10^5 \le \Ra \le 10^{11}$ in compressed
119: gas (air) at room temperature ($\Pr = 0.7$),
120: but still at fixed lateral position.
121: \cite{lui98} measured the lateral dependence of
122: $\lambda_\theta^{sl} (*)$
123: on the
124: positions $x$ (in the mean LSC direction) and $y$ (perpendicular
125: to the preferred LSC direction) in a cylindrical water filled RB cell with $\Gamma = 1$ in the regime
126: $\Ra = 2 \cdot 10^8 - 2\cdot 10^{10}$. They report variations of the BL width by nearly
127: a factor of 2, depending on the lateral location where the profile was measured. \cite{wan03} found
128: similar results for a cubic cell.
129: Even the scaling exponent of $\lambda_\theta^{sl} (*)$ with
130: $\Ra$ depends on the position and varies between $-0.35$ and
131: $-0.28$. \cite{lui98} found the thermal BL to be thinnest
132: close to the center of the plates; there
133: $2\lambda_\theta^{sl} (0,0)/L = (0.23\pm 0.02) \Ra^{-0.285\pm 0.004}$
134: holds.
135:
136:
137: One cannot draw conclusions about the dependence of Nu on Ra from such measurements of local
138: $\lambda_\theta^{sl}(*)$ at a particular lateral location. It is the length scale resulting
139: from the {\it laterally averaged} BL temperature profile
140: which is connected with the Nusselt number through the exact relation
141: \be
142: \Nu = {|\partial_z \left< \theta (x,y, z=0 \ \hbox{or} \, \, z=L)\right>_A |
143: \over {\Delta L^{-1}} } =
144: { L \over 2 \lambda_\theta^{sl} }
145: \label{nu-lambda}
146: \ee
147: which follows from equation (\ref{nu}).
148:
149:
150: A different definition for a thermal boundary-layer thickness is the
151: one employed by classical (laterally homogeneous) laminar BL theories
152: (\cite{pra04,bla08},
153: see also \cite{mek61,sch00,cow01}).
154: This thickness, known as $\lambda_\theta^{99\%}$, is defined
155: as the vertical
156: distance from the plate to the point where $99\%$ of
157: the temperature
158: difference between plate and
159: mean center temperature is achieved.
160: E.g.\ for the bottom plate $\lambda_\theta^{99\%}$ is the distance
161: where the temperature reaches $T = T_b - 0.99 \Delta_b$.
162: This definition is in analogy to the common definition of the thickness $\delta$ of the
163: kinetic BL, defined by the distance where, say, 99\% of the maximum bulk velocity is achieved.
164: The profile-based thickness $\lambda_\theta^{99\%}$ can be calculated within
165: the truly laminar Prandtl-Blasius BL theory, which implies lateral homogeneity and thus yields local
166: and area-averaged widths that are the same. One obtains
167: \be
168: \frac{\lambda_\theta^{99\%}}{L} = \frac{C(\Pr)}{\Re^{1/2} \Pr^{1/3}}
169: \label{pohlhausen-thickness}
170: \ee
171: with a function $C(\Pr)$ given by \cite{mek61}. For large Prandtl numbers
172: $C(\Pr)$ becomes constant, whereas for small $\Pr$ one finds $C(\Pr) \propto \Pr^{-1/6}$.
173: We note that $\lambda_\theta^{99\%}$ can display different
174: scaling behavior than $\lambda_\theta^{sl}$, both as a function
175: of Ra and of Pr, depending on the parameter-space regime. To our knowledge this issue has not yet been
176: explored systematically, neither for $\lambda_\theta^{99\%}$, nor for its local counterpart $\lambda_{\theta}^{99\%}(*)$
177: which depends on the lateral position $*$ and refers to the local, only
178: time-averaged, profiles.
179:
180: A third way to define a thermal BL thickness, suggested by \cite{bel94}, is to take the
181: position of the maximal temperature fluctuations at the edge of the thermal BL.
182: We call the respective BL thicknesses defined in this way $\lambda_\theta^\sigma$ and $\lambda_\theta^\sigma(*)$.
183: \cite{pui07} show that thermal BL thicknesses defined in this way
184: and measured in the Barrel of Ilmenau using air at one atmosphere as the fluid, behave differently from $\lambda_\theta^{sl}$.
185: However, since BL thicknesses are determined largely by diffusive processes,
186: it cannot be ruled out that a mixture like air (where both mass diffusion
187: and heat diffusion play a role and where marginal stability depends both
188: on concentration and temperature gradients) might in this respect behave
189: differently from a pure fluid.
190:
191:
192: \begin{figure}
193: \begin{center}
194: %\includegraphics*[height=5.5cm]{Figs/bl_thess_dupuits.eps}
195: \includegraphics*[height=5.5cm]{Figs/Fig9_a_new.eps}
196: \vspace{0.5cm}
197: %\includegraphics*[height=5.5cm]{Figs/Fig9b_1.eps}
198: \includegraphics*[height=5.5cm]{Figs/Fig9_b_new.eps}
199: \caption{
200: (a) Local, only time-averaged, temperature and temperature-fluctuation profiles
201: at $\Ra = 7.7 \cdot 10^{11}$ and $\Gamma = 1.13$ as functions of the height
202: z (normalized by $\lambda_\theta^{sl}$),
203: measured below the center of the top plate.
204: The fluid is air at atmospheric pressure, the cell height is 6.3 m.
205: From \cite{pui07}.
206: (b)
207: Rescaled mean velocity profiles $U(z)$ above the center of the plate
208: for
209: $\Ra = 1.25 \cdot 10^9$ (shortest data set) up to
210: $\Ra = 2.02 \cdot 10^{10}$ (longest data set) for RB flow in water.
211: $u$ is normalized by $u_{max}$ and $z$ by the
212: kinetic BL thickness $\lambda_u^{sl}(*)$.
213: %Furthermore: We might wish to use smaller case $u$ and $u_{max}$ also in the figure, since we always used lower case letters else in the review.
214: Figure taken from \cite{sun08}.
215: }
216: \label{bl-profiles}
217: \end{center}
218: \end{figure}
219:
220:
221:
222: \begin{figure}
223: \begin{center}
224: %\includegraphics*[height=5.5cm]{Figs/Fig10_a.eps}
225: \includegraphics*[height=5.5cm]{Figs/Fig10_a_new.eps}
226: \vspace*{0.9cm}
227: %\includegraphics*[height=5.5cm]{Figs/Fig10_b.eps}
228: \includegraphics*[height=5.5cm]{Figs/Fig10_b_new.eps}
229: \caption{
230: Scaling of the kinetic BL thicknesses $\lambda_u^{sl}(*)$ and $\lambda_u^m(*)$
231: as functions of (a) Ra and of (b) Re, taken above the center
232: of the lower plate.
233: Taken from \cite{sun08}.
234: In (a) also the scaling of the thermal BL thickness $\lambda_\theta^{sl}
235: (*)$ is shown.
236: It has been deduced from the temperature profile above the center
237: of the lower plate.
238: The lines are power-law fits.
239: }
240: \label{bl-thicknesses}
241: \end{center}
242: \end{figure}
243:
244:
245:
246: Profile measurements, and thus the determination of BL thicknesses,
247: are obviously less difficult if the absolute length scale $L$ is larger. That is
248: why the about 6.5 m high barrel of Ilmenau is particularly suited for these purposes.
249: \cite{pui07} measured local time-averaged temperature profiles
250: up to $\Ra = 7.7 \cdot 10^{11}$, for an
251: example see figure \ref{bl-profiles}a. Outside a small linear regime near the plate
252: they obtain a power law $\theta (z) \sim z^\alpha$ with $\alpha \approx 0.5$,
253: only weakly dependent on Ra and $\Gamma$.
254: For even larger $z$ the temperature saturated at its bulk value. The 1/2 power law for intermediate heights
255: supports neither the logarithmic scaling laws for turbulent BL theories nor the temperature profile
256: for a one-way coupled
257: temperature field (i.e., taking the temperature as a passive scalar)
258: predicted by the
259: (laminar) Prandtl-Blasius approach. However,
260: the power law $\lambda_\theta^{sl}(*)/L \propto Ra^{-0.254}$ found by \cite{pui07} below the center
261: of the top plate is not inconsistent with what follows from the scaling-wise laminar Prandtl-Blasius approach.
262: \cite{sun08} obtained $\lambda_\theta^{sl}(*)/L \propto Ra^{-0.33\pm 0.05}$
263: from temperature-profile measurements above the center of the plate
264: in water with Ra up to $ 2 \cdot 10^{10}$.
265:
266:
267: Thermal BL thicknesses also have been determined from DNS. For a $\Gamma = 1$
268: RB cell \cite{ver99} obtained for the area-averaged width
269: $\lambda_\theta^{sl} = 3.1 Ra^{-0.29}$ for $Pr= 0.7$ over the range $5\cdot 10^5 \le \Ra \le 2\cdot 10^7$
270: and $\lambda_\theta^{sl} = 2.8 Ra^{-0.25}$ for $Pr= 0.022$ in the regime $5\cdot 10^4 \le \Ra \le 10^6$.
271: When going towards the larger Ra regime $2\cdot 10^6 \le \Ra \le 2\cdot 10^{11}$
272: \cite{ver03} find $\lambda_\theta^{\sigma} \sim Ra^{-0.31}$ for $Pr= 0.7$ and $\Gamma = 1/2$.
273: For even larger $\Ra$ up to $ 2\cdot 10^{14}$ \cite{ver08} report
274: $\lambda_\theta^{\sigma} \sim Ra^{-1/3}$.
275:
276:
277:
278: The profiles of the rms temperature fluctuations are much less settled,
279: see e.g.\ \cite{lui98,fer02,wan03,pui07,sun08}. This issue will not be discussed in the present review.
280:
281:
282: As stated already in section \ref{nuregl}, thermal {\it plumes} can be
283: viewed as detached pieces of the thermal BL.
284: Since their discovery in RB convection by \cite{zoc90} they have
285: been studied so intensively, -- experimentally, numerically,
286: and theoretically -- that a review of its own on this subject
287: would be justified. A number of recent papers were devoted to them
288: (\cite{the98,the00,zho02,bre04,fun04,har04,xi04,put05,put05b,shi06,zho07,put08,shi08}).
289: Their nature was discussed most recently by \cite{FBA08}. Here we only make a few remarks. They seem to
290: originate as one-dimensional excitations of the marginally stable BL. As they are born out of the BL,
291: they become oriented by the LSC with their long axis in the direction of the flow, and then are swept
292: by the LSC toward the side wall. Along their way they separate from the BL and progress vertically near
293: the wall, developing their famous mushroom top (\cite{zoc90,xi04,put05,zho07}) in this process. Although
294: they are often referred to as `sheet-like", implying an extension in two spatial dimensions,
295: the term ``line-like" would more appropriately describe their length scale of order $L$ in one dimension
296: and of order the thermal boundary-layer thickness
297: in the other two.
298:
299:
300:
301:
302: \subsection{Kinetic boundary layers}
303:
304: The kinetic BL thicknesses $\lambda_u$ are determined from the velocity profiles or the velocity-fluctuation profiles. Also $\lambda_u $ can be defined in various ways,
305: starting either from time- and area-averaged quantities or only locally
306: time-averaged ones.
307: An additional alternative here is to average the
308: velocity vectors themselves or to take the rms velocities, component wise or the full magnitude.
309: Given the velocity profile of interest, its thickness may be defined
310: (i) via the slope of the velocity profile in the $z$-direction (for the BL
311: above or below the bottom or top plate)
312: at a given position $*$ on the plate,
313: known as $\lambda_u^{sl}(*)$,
314: (ii) through the distance $\lambda_u^m(*)$ to the local maximal mean velocity,
315: (iii) through the distance $\lambda_u^\sigma(*)$ to the maximal
316: velocity rms fluctuations, etc. Again, all these quantities depend on the lateral position $* = (x,y)$,
317: unless area-averages are considered.
318:
319: The first direct systematic measurements of velocity profiles in RB convection
320: as a function of Ra where done by \cite{xin96} and \cite{xin97}, using a novel light-scattering technique
321: in a cylindrical cell. They found $\lambda_u^{sl}(*)/L \sim Ra^{-0.16}$
322: from the velocity profile above the center of the lower plate.
323: At the side-walls at half height as functions of $x$ instead of $z$ they measured $\lambda_u^{sl}(*)/L \sim Ra^{-0.25}$.
324: \cite{qiu98,qiu98b} extended these measurements to convection in cubic cells, finding the same scaling
325: exponents as in the cylindrical case. By using various organic liquids, \cite{lam02} could also explore
326: the Pr dependence over the range $6 \le \Pr \le 1027$
327: and $2\cdot 10^8 \le \Ra \le 2 \cdot 10^{10}$, finding
328: $\lambda_u^{sl}(*)/L \sim \Pr^{0.24} \Ra^{-0.16}$ above the
329: center plate position. The small $\Ra$-exponent of the effective
330: power law is
331: remarkable, as for a Prandtl-Blasius area-averaged BL profile one would expect an exponent of
332: $-0.25$, just as measured at the side-walls, roughly corresponding to
333: the Prandtl-Blasius scaling $\lambda_u^{sl}/L \sim \Re^{-1/2}$ with (approximately)
334: $\Re \sim \Ra^{1/2}$. We will come back to this point later.
335: Note that all these measurements were still done above a single
336: position over the plate.
337:
338: In 2003, particle image velocimetry (PIV) measurements
339: revolutionarized the experimental analysis of the velocity field
340: in thermal convection (\cite{xia03,sun05,sun05a,sun06}),
341: including the analysis of the kinetic BLs (\cite{sun08}).
342: This allowed not only for the direct identification of various Reynolds numbers, cf.\ Sect. \ref{nurere},
343: but also for that of various kinetic BL thicknesses such as $\lambda_u^{sl}(*)$,
344: $\lambda_u^{m}(*)$, or $\lambda_u^{\sigma}(*)$, even at different positions $*$ in the cell.
345:
346: For $10^9 \le \Ra \le 10^{10}$ and $\Pr = 4.3$ in a rectangular cell
347: with $\Gamma = 1$, \cite{sun08} obtained time-averaged velocity profiles
348: in the center above the bottom plate. They found that if the vertical lengths are rescaled by $\lambda_u^{sl}(*)$
349: and the velocities by the maximal velocity, then the shape of the local velocity profile does not depend on
350: $\Ra$; in this sense it is universal in that regime, see figure \ref{bl-profiles}b.
351: Moreover, \cite{sun08} find $\lambda_u^{sl}(*)/L\sim \lambda_u^{m}(*)/L \sim \Ra^{-0.27\pm 0.01}$
352: (see figure \ref{bl-thicknesses}a), in contrast to the much weaker Ra dependence reported earlier
353: (\cite{xin96,xin97,qiu98,qiu98b}. The origin of this discrepancy is not clear at this point. However,
354: we note that the newly found scaling
355: $\lambda_u^{sl}(*)/L\sim \lambda_u^{m}(*)/L \sim \Ra^{-0.27\pm 0.01}$
356: is consistent with the Prandtl-Blasius expectation. As a double check,
357: \cite{sun08} plotted $\lambda_u^{sl}(*)/L$ and $\lambda_u^{m}(*)/L$ against
358: the independently determined Reynolds number, finding the Prandtl-Blasius
359: scaling $\lambda_u^{sl}(*)/L \sim \lambda_u^{m}(*)/L \sim Re^{-0.50 \pm 0.03}$, see
360: figure \ref{bl-thicknesses}b.
361: In numerical simulations of RB flow for $\Pr = 1$ and $\Gamma = 1$
362: \cite{bre04} find $\lambda_u^{sl}/L \sim \Re^{0.44\pm 0.02}$,
363: but a weaker Re-dependence for $\lambda_u^{m}/L$.
364: Numerical simulations by \cite{ver99}
365: for the same $\Gamma = 1$ and $\Pr =0.7$
366: gave an area-averaged profile thickness exponent consistent with the Prandtl-Blasius
367: BL theory, namely $\lambda_u^{sl} /L = 0.95 Ra^{-0.23}$.
368: For lower $\Pr = 0.022$ \cite{ver99} obtained
369: $\lambda_u^{sl}/L = 0.1 Ra^{-0.18}$.
370:
371:
372:
373: The PIV BL study of \cite{sun08} also yielded the wall shear
374: stress $\tau_w$, the viscous sublayer lengthscale $y_w = \nu(z=0)/u_\tau$,
375: the skin-friction velocity $u_\tau = (\tau_w / \rho )^{1/2}$,
376: and the skin-friction coefficient $c_f = \tau_w / (\rho U_{max}^2)$.
377: (All these quantities have been measured above the plate center, but here
378: for simplicity we suppress the $*=(x,y)$-dependence in the notation.)
379: They found
380: $\tau_w/(\rho \nu^2 /L^2) \sim Re^{1.55}$,
381: $y_w/L \sim Re^{-0.91}$,
382: $u_\tau/(\nu/L) \sim Re^{0.80}$,
383: and
384: $c_f \sim Re^{-0.34}$.
385: The respective Prandtl-Blasius scaling exponents for a laminar BL over
386: a flat plate are
387: $3/2$, $-1$, $3/4$, and $-1/2$. Apart from the last one all are in fair
388: agreement with the measurements. It is hard to judge the importance
389: of the discrepancy between $-0.34$ and $-0.50$ for the scaling of
390: the skin-friction coefficient, in particular because the last three
391: data points around $\Re \approx 10^4$ show a steeper dependence
392: than $-0.34$, see figure 22d of \cite{sun08}. Moreover,
393: in the numerical simulations of \cite{ver03}
394: ($\Gamma = 1/2$, $\Ra$ up to $2\cdot 10^{11}$)
395: the friction coefficient $c_f$ does show a Reynolds-number scaling-exponent consistent with $-1/2$.
396: (That work did not find any scaling
397: for $\lambda_u^\sigma$ as function of Ra, presumably because of a
398: reorganization of the flow in the $\Gamma = 1/2$ cell with increasing Ra.)
399: Also in the hitherto largest numerical RB simulations of
400: \cite{ama05} $c_f \sim Re^{-1/2}$ is found, see figure 6b of that paper.
401: Next, also the experimental analysis of
402: \cite{cha01} finds a $-1/2$ scaling of a surrogate of the friction factor with
403: $\Re$ up to $\Ra = 10^{11}$. However, at $\Ra = 10^{11}$, corresponding
404: to $\Re = 8 \cdot 10^4$, the data of \cite{cha01} give evidence for a transition
405: towards a weaker dependence. \cite{ver03} find the same transition
406: in this friction factor surrogate, but not in the actual friction factor
407: $c_f$.
408:
409: Finally, we would like to mention the interesting numerical study by
410: \cite{yu07}, who analyse the heat transfer and boundary-layer thicknesses
411: in {\it laminar} and steady convection with $\Ra$ up to $10^8$,
412: finding basically the same Ra-scaling laws for
413: Nu, $\lambda_u$, and $\lambda_\theta$ as in classical RB convection.
414: Their conclusion is that turbulence does not play a decisive role for
415: the heat transfer.
416:
417:
418: In summary, it is fair to state that, although some of the various velocity BL profiles
419: differ from the idealized
420: Prandtl-Blasius truly laminar profile due to the permanently ongoing plume emission, the
421: {\it scaling} of the kinetic BL thickness with Ra, Re, and Pr
422: is consistent with the laminar Prandtl-Blasius
423: theory at least up to $\Ra = 10^{11}$, but presumably also beyond.
424: There is no indication of any transition towards a different kinetic BL
425: thickness-scaling even at the largest Ra realized up to now.
426: Moreover, for $\Pr \gtwid 1$ the thermal BL is nested in the kinetic one,
427: whereas for $\Pr \ltwid 1$ it is the other way round.
428:
429:
430: The scaling of the rms velocity fluctuations (see e.g. \cite{sun08,pui07b} for recent work on this issue)
431: will not be discussed in this review.
432:
433:
434:
435: