1: \documentclass[a4paper,11pt,leqno]{article}
2: \usepackage{epsfig, graphics, amssymb, amsfonts, euscript ,float,graphicx,color}
3: \hoffset=-20mm \voffset= -20mm \textwidth 17cm \textheight 22cm
4: \def\today{\number\day\space\ifcase\month\or
5: January\or February\or March\or April\or May\or June\or July\or
6: August\or September\or October\or November\or December\fi
7: \space\number\year}
8: \def\bcdot{\mathbin{\mbox{\boldmath$\cdot$}}}
9: \def\erf{\mathop{\rm erf}\nolimits}
10: \def\sign{\mathop{\rm sign}\nolimits}
11: \makeatletter
12: \def\@begintheorem#1#2{\trivlist \item[\hskip \labelsep{\sc #1\ #2.}]\sl}
13: \def\@opargbegintheorem#1#2#3{\trivlist
14: \item[\hskip \labelsep{\sc #1\ #2\ (#3).}]\sl}
15: \def\@endtheorem{\endtrivlist}
16: \def\@sect#1#2#3#4#5#6[#7]#8{\ifnum #2>\c@secnumdepth
17: \let\@svsec\@empty\else
18: \refstepcounter{#1}\edef\@svsec{\csname the#1\endcsname.\hskip
19: 0.5em}\fi \@tempskipa #5\relax \ifdim \@tempskipa>\z@
20: \begingroup #6\relax
21: \@hangfrom{\hskip #3\relax\@svsec}{\interlinepenalty \@M #8\par}
22: \endgroup \csname #1mark\endcsname{#7}\addcontentsline
23: {toc}{#1}{\ifnum #2>\c@secnumdepth \else
24: \protect\numberline{\csname the#1\endcsname}\fi #7}\else
25: \def\@svsechd{#6\hskip #3\relax\@svsec #8.\csname #1mark\endcsname
26: {#7}\addcontentsline {toc}{#1}{\ifnum #2>\c@secnumdepth \else
27: \protect\numberline{\csname the#1\endcsname}\fi #7}}\fi
28: \@xsect{#5}}
29: \makeatother \pagenumbering{arabic}
30: \newtheorem{Th}{Theorem}[section]
31: \newtheorem{Prop}{Proposition}[section]
32: \newtheorem{Lm}{Lemma}[section]
33: \newtheorem{Cor}{Corollary}[section]
34: \newtheorem{Rm}{Remark}[section]
35: \newtheorem{Def}{Definition}[section]
36: \newtheorem{theorem}{Theorem}
37: \newtheorem{proposition}{Proposition}
38: \newtheorem{lemma}{Lemma}
39: \newtheorem{corollary}{Corollary}
40: \newtheorem{definition}{Definition}
41: \newtheorem{remark}{Remark}
42: \renewcommand{\theequation}{\thesection.\arabic{equation}}
43: \renewcommand{\thetheorem}{\thesection.\arabic{theorem}}
44: \renewcommand{\thelemma}{\thesection.\arabic{lemma}}
45: \renewcommand{\thecorollary}{\thesection.\arabic{corollary}}
46: \renewcommand{\thedefinition}{\thesection.\arabic{definition}}
47: \renewcommand{\theremark}{\thesection.\arabic{remark}}
48: \newenvironment{Proof}{\removelastskip\vskip12pt plus 1pt \noindent\sc
49: Proof.\quad\rm}{\hspace*{\fill}$\blacksquare$}
50: \newenvironment{Pf}{\removelastskip\vskip12pt plus 1pt \noindent
51: \rm}{\hspace*{\fill}$\blacksquare$}
52: \setlength{\leftmargini}{0pt} \setlength{\leftmarginii}{0pt}
53: \newtheorem{Thf}{\sc Th\'eor\`eme}[section]
54: \newtheorem{Propf}{\sc Proposition}[section]
55: \newtheorem{Lmf}{\sc Lemme}[section]
56: \newtheorem{Corf}{\sc Corollaire}[section]
57: \newtheorem{Remf}{\sc Remarque}[section]
58: \newtheorem{Deff}{\sc D\'efinition}[section]
59: \renewcommand{\theequation}{\thesection.\arabic{equation}}
60: \def\isdef{\buildrel{\rm def}\over=}
61: \def\supp{\mathop{\hbox{supp}}}
62: \newcommand{\be}{\begin{equation}}
63: \newcommand{\bT}{\begin{array}}
64: \newcommand{\eT}{\end{array}}
65: \newcommand{\de}{\end{equation}}
66: \newcommand{\ee}{\end{equation}}
67: \def\R{\mathbb{R}}
68: \def\s{\sigma}
69: \def\ub{\bar u}
70: \def\ut{u_t}
71: \def\t{\gamma}
72: \def\ux{u_x}
73: \def\uz{u_0}
74: \def\text#1{{\quad \hbox{#1} \quad}}
75: \def\QT{Q_T}
76: \def\ds{\displaystyle}
77: \def\sec{{\prime\prime}}
78: \def\tr{\prime\prime\prime}
79: \begin{document}
80: \title{\bf On similarity and pseudo-similarity solutions of Falkner-Skan boundary layers}
81: \author{
82: \bf M.~\textsc{\bf Guedda} and \bf Z.~\textsc{\bf Hammouch}\footnote{Corresponding author. E-mail: zakia.hammouch@u-picardie.fr} \\
83: {\footnotesize \it LAMFA, CNRS UMR 6140, Universit\'e de Picardie
84: Jules Verne,
85: }\\
86: {\footnotesize \it Facult\'e de Math\'ematiques et d'Informatique,
87: 33, rue Saint-Leu 80039 Amiens, France} }
88: \date{}
89: \maketitle
90:
91: \noindent \rule{17cm}{0.02cm} {\bf Abstract}\\
92:
93: {\footnotesize The present work deals with the two-dimensional
94: incompressible, laminar, steady-state boundary layer equations.
95: First, we determine a family of velocity distributions outside
96: the boundary layer such that these problems may have similarity
97: solutions. Then, we examen in detail new exact solutions, called
98: {\it Pseudo--similarity}, where the external velocity varies
99: inversely--linear with the distance along the surface $ (u_e(x) =
100: u_\infty x^{-1}). $ The analysis shows that solutions exist only
101: for a lateral suction. For specified conditions, we establish the
102: existence of an infinite number of solutions, including monotonic
103: solutions and solutions which oscillate an infinite number of
104: times and tend to a certain limit. The properties of solutions
105: depend on
106: the suction parameter. Furthermore, making use of the fourth--order Runge--Kutta scheme together with the shooting method, numerical solutions are obtained.}\\
107: \noindent{\scriptsize {\it \bf Keywords:}\quad
108: Similarity solutions, Pseudo-similarity solutions, Falkner-Skan
109: problem, Boundary Layer, Stretching surfaces.\newline
110: {\it \bf PACS
111: numbers:}\quad 47.15, 47.27 Te\\ }
112: \rule{17cm}{0.02cm}\\
113:
114: %---------------------------------------------------------------------introduction----------------------------------------------
115: \section{Introduction}\label{Sect.1}
116: \setcounter{equation}{0} \setcounter{theorem}{0}
117: \setcounter{lemma}{0} \setcounter{remark}{0}
118: \setcounter{corollary}{0}
119:
120: \qquad In this paper we are concerned with the classical
121: two-dimensional laminar incompressible boundary layer flow past a
122: wedge or a flat plate \cite{Sch}. For the first approximation, the
123: model is described by the Prandlt equations or the boundary
124: layer equations
125: \begin{equation}\label{eq:Prandtl}
126: u\partial_x u + v\partial_y u= u_e\partial_x
127: u_e+\nu\partial^2_{yy} u, \quad \partial_x u + \partial_y v = 0,
128: \end{equation}
129: where $(x,y)$ denote the usual orthogonal Cartesian coordinates
130: parallel and normal to the boundary $ y = 0 $ (the wall), $ u$
131: and $v $ are the corresponding
132: velocity components, and the constant $ \nu > 0 $ is the kinematic-viscosity. The function $ u_e $ is a given exterior streaming velocity flow
133: which is assumed throughout the paper to be nonnegative function of the single variable $x; u_e = u_e(x),$ and is such
134: that $ u(x,y) $ tends to $ u_e(x) $ as $ y\to \infty.$
135: Equations (\ref{eq:Prandtl}) can be written in the form
136: \begin{equation}
137: \partial_y\psi\partial_{xy}^2\psi - \partial_{x}\psi\partial_{yy}^2\psi = u_e\partial_x u_e+\nu\partial^3_{yyy}\psi,
138: \end{equation}
139: where $ \psi $ is the well--known stream function defined by $ u
140: = \ds\partial_{y}\psi, v = -\partial_{x}\psi.$
141:
142: This problem with appropriate external velocity flow has been the
143: main focus of studies of particular exact solutions. Research on
144: this subject
145: has a long history, which dates to the
146: pioneering works by Blasius \cite{Blas} and Falkner and Skan
147: \cite{FS} in which the external velocity is given by
148: \begin{equation}
149: u_e(x) = u_\infty x^m \quad (u_\infty > 0).\end{equation}
150: Their investigations lead to solutions to (1.2) in the form
151: \begin{equation}\label{eq:similarity}
152: \psi(x,y) = ax^{\alpha} f(byx^{-\beta}), \ a, b > 0,
153: \end{equation}
154: where
155: \begin{equation}
156: \alpha = \frac{m+1}{2}\ \mbox{and}\ \beta =
157: \frac{1-m}{2}.\end{equation} The equation for $ f $ is the
158: well-known equation obtained by Falkner-Skan \cite{FS}
159: \begin{equation}\label{eq:FS}
160: f^{\sec\prime}+ \frac{m+1}{2}ff^\sec + m({1-f^\prime}^2)=0\quad
161: \mbox{ on}\quad (0,\infty),
162: \end{equation}
163: or, if $ m > -1,$
164: \begin{equation}\label{eq:FSgamma}
165: f^{\sec\prime}+ ff^\sec + \sigma({1-f^\prime}^2)=0\quad \mbox{
166: on}\quad (0,\infty),
167: \end{equation}
168: where
169: \[ \sigma= \frac{2m}{m+1}.\]
170: Such equations occur at wedge flows \cite[ p. 170]{Sch}. These
171: equations have received considerable attention in the literature.
172: We refer the reader to the works of Rosenhead \cite{Rose},
173: Schlichting and Gersten \cite{Sch}, Weyl \cite{Weyl} and Coppel
174: \cite{Cop11} and the references therein. Note that from (1.4) it
175: is easily verified that,
176: \[ \psi(x,y) = \left(\frac{x}{x_0}\right)^{\alpha}\psi\left(x_0,y\left(\frac{x}{x_0}\right)^{-\beta}\right).\]
177: This means that a solution $ \psi(x,y) $ for $ y $ fixed is
178: similar to the solution $ \psi(x_0,y) $ at a certain $ x_0.$
179: This
180: solution is called invariant or similarity solution and the
181: function $ f
182: $ is called the shape function or the dimensionless stream function.\\
183: The broad goals of this paper is to study equation (1.7) when
184: taking the limit $\sigma \to -\infty.$ This limit case,
185: corresponding
186: to $ m = -1,$ is that of flow in a two-dimensional divergent channel (or sink flow) \cite[p. 170]{Sch}. We prove that transformation (1.4) is much too restrictive; that
187: is problem (1.1) has no similarity solution and we shall see, by
188: rigouros arguments, that the term $ \gamma log(x) $ must be
189: added to the expression (1.4) and the surface must be permeable
190: with suction to obtain new exact solutions which are not
191: similarity. In addition, we shall prove that there is an
192: infinite number of solutions. These results are given in Section
193: 3. Before this analysis, we shall identify, in Section 2, external
194: flows, such that problem (1.2) may have similarity solutions. The
195: main result of this section indicates that problem (1.2) has
196: solutions under the form (1.4) if the external flow is of the
197: power--law type (1.3).
198:
199: \section{Similarity solutions}
200: \setcounter{equation}{0} \setcounter{theorem}{0}
201: \setcounter{lemma}{0} \setcounter{remark}{0}
202: \setcounter{corollary}{0} \qquad In this section we shall obtain
203: external flows such that the partial differential equation (1.2)
204: accompanied by the boundary condition
205: \begin{equation}
206: \lim_{y\to \infty}\partial_y\psi(x,y) = \partial_y\psi(x,\infty) = u_e(x),
207: \end{equation}
208: has a solution under the form (1.4), where $ \alpha+\beta = 1.$
209:
210: The main problems, arising in the study of similarity solutions,
211: are related to the existence of the exponents $ \alpha $ and $
212: \beta $ and to the rigorous study of the ordinary differential
213: equation satisfied by the profile $ f$ which is, in general,
214: nonlinear. For the layer equation (1.2), the classical approach
215: for identifying $ \alpha $ and $ \beta $ is the scaling and
216: transformation group \cite{Bar}. The essential idea is to seek $
217: a $ and $ b $ such that if $ \psi $ satisfies (1.2) the new
218: function $ \psi_\kappa(x,y) = \kappa^a\psi(\kappa^b x,
219: \kappa y) $ is also a solution. \\
220: Let $ \psi $ be a stream-function to (1.2) defined by (1.4) where
221: $ \alpha + \beta = 1.$ Assume first that $ \beta\not=0.$ We
222: choose $ a = -\frac{\alpha}{\beta},\ b = \frac{1}{\beta},$ and
223: define $\psi_\kappa(x,y) = \kappa^a\psi\left(\kappa^bx,\kappa
224: y\right).$ Hence $ a + b = 1, \psi\equiv \psi_\kappa $ and
225: \[ L(\psi_\kappa)(x,y) = \kappa^{a+3}L(\psi)(\kappa^bx,\kappa y )\] for any $ \kappa > 0,$
226: where $ L $ is the operator defined by
227: \[ L(\psi) =\partial_y\psi\partial_{xy}^2\psi - \partial_{x}\psi\partial_{yy}^2\psi-\nu\partial^3_{yyy}\psi.\]
228: According to equation (1.2) we deduce
229: \[ h(x)=\kappa^{a+3}h(\kappa^bx),\]
230: where $ h(x) = u_e(x)\partial_x u_e(x).$ In particular, for fixed $
231: x_0 > 0 $
232: \[ h(\kappa^bx_0)=\kappa^{-(a+3)}h(x_0).\]
233: Setting $ x = \kappa^bx_0 $ we infer
234: \[h(x)=x^{-\frac{a+3}{b}}x_0^{\frac{a+3}{b}}h(x_0).\]
235: Solving the equation
236: \begin{equation} u_e\frac{d u_e}{dx} = x^{-\frac{a+3}{b}}x_0^{\frac{a+3}{b}}h(x_0)\end{equation} yields us
237: \begin{equation}\label{eq:UE}
238: u_e^2(x) = {c_1x^{2m}+c_2},\end{equation}
239: for all $ x > 0, $ where $ m = \alpha-\beta$ and $ c_1 $ and $
240: c_2 $ are constants, for $\beta\not=0$, since $ -\frac{a+3}{b}
241: +1=2(\alpha-\beta). $\\ For $ \beta = 0, $ hence $ \alpha =m =
242: 1,$ the new function
243: \[ \psi_\kappa(x,y)= \kappa^a\psi(\kappa^{-a}x,y),\]
244: for any fixed $ a \not= 0, $ is equivalent to $ \psi $ and
245: satisfies
246: \[ L(\psi_\kappa)(x,y) = \kappa^{a}L(\psi)(\kappa^{-a}x,y )\] for any $ \kappa > 0.$ Arguing as in the case $ \beta \not=0 $ one arrives at (2.3) with $ m
247: = 1.$ Next, because $
248: \lim_{y\to\infty}x^{2m}f^\prime(yx^{-\beta})^2 = c_1x^{2m}+ c_2,$
249: the function $ {f^\prime}^2 $ has a finite limit at infinity,
250: which is unique and is given by $ c_1 + c_2x^{-2m}.$ This is
251: acceptable only for $ c_2 = 0.$
252:
253: The above result indicates, in
254: particular, that for a prescribed external velocity satisfying
255: (2.3)
256: the real numbers
257: $ \alpha $ and $ \beta $ are given by (1.5) and condition (2.3)
258: is necessary and sufficient to obtain similarity solution under
259: the form (1.4) where $ \alpha+\beta = 1.$ However, for a
260: general external velocity, it is possible to obtain an exact solution which is not similarity solution as it is seen in
261: \cite{LCB}. In this paper the authors considered
262: \[ u_e(x) = c_1x^{1/3} + c_2x^{-1/3}. \]
263: A stream--function $ \psi $ is given by
264: \begin{equation}
265: \psi(x,y)=x^{2/3} f(yx^{-1/3}) + cx^{-\frac{1}{3}}y,\quad c =
266: const,\end{equation} where $ f $ is a solution of
267: \begin{equation}
268: \nu f^{\sec\prime}\ +\ \alpha ff^\sec
269: -(\alpha-\beta){f^\prime}^2=\xi,\quad \xi = const.
270: \end{equation}
271:
272:
273: \section{ The pseudo-similarity solutions}
274: \setcounter{equation}{0} \setcounter{theorem}{0}
275: \setcounter{lemma}{0} \setcounter{remark}{0}
276: \setcounter{corollary}{0} \qquad In the present section we
277: restrict the attention to the case $ m = -1 $ and get new
278: solutions to the problem
279: \begin{equation}
280: \partial_y\psi\partial_{xy}^2\psi - \partial_{x}\psi\partial_{yy}^2\psi = \nu\partial_{yyy}^3\psi-u_\infty^2 x^{-3},
281: \end{equation}
282: subject to the boundary conditions
283: \begin{equation}
284: \partial_y\psi(x,0) = u_w x^{-1},\quad \partial_x\psi(x,0) = -v_wx^{-1},\quad \partial_y\psi(x,\infty)=u_\infty x^{-1},
285: \end{equation}
286: where $ v_w $ is a real number ( $v_w>0$ for suction and $v_w<0$
287: for injection), $ u_w $ and $ u_\infty $ are nonnegative and
288: satisfy $ u_w < u_\infty.$ The velocity distribution $ u_e
289: =\frac{u_\infty}{x }$ is found in the case of divergent channel
290: (or sink flow) \cite[p. 170]{Sch}. The analysis of this section
291: is motivated by the work by Magyari, Pop and Keller \cite{MPK}
292: concerning a boundary--layer flow, over a permeable continuous
293: plane surface, where the $x$-component velocity tends to zero for
294: $ y $ large ($u_\infty = 0$). In \cite{MPK} the authors showed
295: that if $ m = -1 $ problem (1.2),(2.1) has no solution in the
296: usual form (1.4). For $ u_\infty\not=0$ and according to Section
297: 2, the function $ \psi $ can be written as
298: \[ \psi(x,y) = \sqrt{\nu u_\infty}f\left(\sqrt{\frac{u_\infty}{\nu}}yx^{-1}\right).\]
299: Since $ v_w = \frac{m+1}{2}\sqrt{\nu u_\infty}f(0) $ (see the
300: Appendix) we deduce $ v_w = 0 $ for $ m=-1 $ and the following
301: oridinary differential equation for $ f $
302: \begin{equation}
303: \left\{\begin{array}{lll}
304: f^{\sec\prime} +{f^\prime}^2-1 = 0, \\
305: \\
306: f^\prime(0) = \zeta, f^\prime(\infty) = 1,
307: \end{array}\right.\end{equation}
308: where $ \zeta = \frac{u_w}{u_\infty} $ is in the interval $ [0,1)
309: $ and $ f(0) $ can be any real number. Since this problem does not
310: contain $ f $ it is convenient to study the equation satisfied by
311: $ \theta = f^\prime; $ that is
312: \begin{equation}
313: \left\{\begin{array}{lll}
314: \theta^{\sec} +{\theta}^2-1 = 0, \\
315: \\
316: \theta(0) = \zeta, \theta(\infty) = 1,
317: \end{array}\right.\end{equation}
318: \\ The stability of equilibrium point $ (1,0) $ of (3.4) cannot
319: be determined from the linearization. To analyze the behavior of
320: the nonlinear equation (3.4)$_1$, we observe that
321: \[ E^\prime(t)=0,\]
322: where $E$ is the Liapunov function given by\\
323: \[ E(t) = \frac{1}{2}\theta^\prime(t)^2 + \frac{1}{3}\theta(t)^3 - \theta(t).\]
324: Then, for some constant $ c,$ the following
325: \[ \theta^{\prime}=\pm\sqrt{2}\left(c+\theta-\frac{1}{3}\theta^3\right)^{1/2},\]
326: holds. The analysis of the algebraic equation of the phase path in
327: the phase plane reveals that the equilibrium point $(1,0)$ is a
328: center.
329: Hence, Problem (3.4) has no solution for any $ \zeta >-1 $ except the trivial one $ \theta = 1$ (see Fig. 3.1). \\
330: Note that if we impose the condition $ \theta(\infty) = -1 $
331: instead of $ \theta(\infty) = 1,$ which is also of physical
332: interest, it is easy to see that, for any $ \zeta \leq 2, $ there
333: exists a unique solution up to translation. This solution
334: satisfies
335: \[ \frac{1}{2}\theta^\prime(t)^2 + \frac{1}{3}\theta(t)^3 - \theta(t)= \frac{2}{3},\]
336: and we find that
337: \[ \theta(t) = 2-3\tanh^2\left[\pm t/\sqrt{2}+{\rm arctanh}\left\{(2-\zeta)/3)^{1/2}\right\} \right].\]
338:
339:
340: \begin{center}
341: \includegraphics[height=2.8in]{falkner1.ps}
342: \end{center}
343: \begin{center}
344: {\footnotesize
345: {Fig. $3.1$} Classification of solutions of $\theta^\sec + \theta^2 -1 = 0 $ according to $ \theta(0) $ and $ \theta^\prime(0).$}
346: \end{center}
347:
348: To obtain exact solutions to (3.1),(3.2), we look for ``{\it pseudo-similarity} " solutions under the
349: form
350: \begin{equation}\label{eq:Similarity}
351: \psi(x,y) = aF(x,byx^{-1}),
352: \end{equation}
353: where $ a = \sqrt{\nu u_\infty} $ and $ b =\sqrt{ \frac{u_\infty}{\nu}}.
354: $ Assuming $ F(x,t) = f(t) + H(x) $ one sees that (cf. Appendix)
355:
356: \begin{equation}
357: \left\{\begin{array}{l}
358: f^{\sec\prime}+ \t f^\sec+ {f^\prime}^2-1=0,\\
359: \\
360: f^\prime(0) = \zeta,\quad f^{\prime}(\infty) = 1,
361: \end{array}\right.
362: \end{equation}
363: and
364: \[ H(x) = \t\log x + c,\]
365: where $ \gamma = v_w\left(\nu u_\infty\right)^{-1/2}$ and $ c $
366: is constant. Without loss of generality we may take $ c = 0,$ since $\psi(x,y) = \sqrt{\nu u_\infty}f(t) + v_w\log(x)+\sqrt{\nu u_{\infty}} c $ satisfies $(3.1)-(3.2)$ for any real number $c$. Note
367: that $ \t $
368: plays the role of suction/injection parameter.\\
369: To study (3.6) it is more convenience to consider the second
370: ordinary differential equation
371: \begin{equation}
372: \left\{\begin{array}{l}
373: \theta^\sec+ \t \theta^\prime+ \theta^2-1=0,\\
374: \\
375: \theta(0) = \zeta,\quad \theta(\infty) = 1,
376: \end{array}\right.
377: \end{equation}
378: where $0 \leq \zeta < 1 $ and $ \t\not=0.$ In fact, the real
379: number $ \t $ will be taken in $ (0,\infty).$ The existence of
380: solutions to (3.7) will be proved by means of shooting method.
381: Hence, the boundary condition at infinity is replaced by the
382: condition $ \theta^\prime(0) = d,$ where $ d $ is a real number.
383: For any $ d $ the new initial--value problem has a unique local
384: solution $ \theta_d $ defined in the maximal interval of the
385: existence, say $(0,T_d), T_d \leq \infty.$ We shall see that for
386: an appropriate $ d $ the solution
387: $ \theta_d $ is global and satisfies
388: \begin{equation} \theta_d(\infty) = 1.\end{equation}
389: A simple analysis in the phase plane shows that problem (3.6) may have solutions only for $\t > 0. $ In fact, the ordinary differential
390: equation in (3.6) is considered as a nonlinear autonomous system
391: in $ \mathbb{R}^2, $ with the unknown $(\theta,\theta^\prime),$
392: mainly
393: \begin{equation}
394: \left\{\begin{array}{l}
395: \theta^\prime=\varphi,\\
396: \\
397: \varphi^\prime =-\t \varphi + 1 - \theta^2,\\
398: \end{array}\right.
399: \end{equation}
400: subject to the boundary condition
401: \begin{equation}
402: \theta(0) = \zeta,\quad \varphi(0) = d.
403: \end{equation}
404: The linear part of the above system at $(1,0)$ is the matrix
405: \[ J=\left(\begin{array}{ccc}
406: 0 & 1\\
407: \\
408: -2 & -\t
409: \end{array}\right). \]
410: The eigenvalues of $ J $ are
411: \[ \lambda_1 = \frac{-\t -\sqrt{\t^2-8}}{2},\quad \lambda_2 = \frac{-\t +\sqrt{\t^2-8}}{2},\]
412: if $ \t \geq \sqrt{8} $ and for $ \vert \t\vert < \sqrt{8}, $
413: \[ \lambda_3 = \frac{-\t -i\sqrt{8-\t^2}}{2},\quad \lambda_4 = \frac{-\t +i\sqrt{8-\t^2}}{2}.\]
414: Therefore, the hyperbolic equilibrium point $(1,0) $ is
415: asymptotically stable if $ \t $ is positive and unstable for negative $ \t
416: $. In particular problem (3.5) has no nontrivial
417: solutions if $ \t < 0.$ If $ \t > 0 $ we deduce from the above
418: that there exists $ \delta > 0 $ such that for any $ d $ and $
419: \zeta $ satisfying $ d^2 + (\zeta - 1)^2 < \delta^2 $ the local
420: solution $ \theta_d $ is global and satisfies (3.8). In the
421: following we construct solution to (3.6) where the condition $
422: d^2 + (\zeta - 1)^2 < \delta^2 $ is not necessarily required.
423: For a mathematical consideration the parameter $ \zeta $ will be
424: taken in $(-1,\sqrt{3}].$ The following result deals with
425: nonnegative values of $ \zeta.$ Let us consider a real number $ d
426: $ such that
427: \begin{equation}
428: d^2\leq 2\zeta(1-\frac{\zeta^2}{3}),
429: \end{equation}
430: where $ 0 \leq \zeta \leq \sqrt{3}. $ We shall see that any local
431: solution of (3.9),(3.10) is global and satisfies (3.8). To this
432: end we consider again the Liapunov function $
433: E(\theta(t),\varphi(t)) = \frac{1}{2}\varphi(t)^2 +
434: \frac{1}{3}\theta^3-\theta.$ Along an orbit we have
435: \[ \frac{d}{dt}E(\theta(t),\varphi(t)) = -\t\varphi(t)^2\leq 0.\]
436: Hence
437: \[ E(\theta_d(t),\theta_d^\prime(t)) \leq E(\zeta,d),\]
438: for any $ t < T_d.$ On the other hand, from (3.9) and (3.10),
439: there exists $ t_0 > 0,$ small, such that $ \theta_d $ is positive
440: on $(0,t_0).$ Assume that $ \theta_d $ vanishes at some $ t_1 >
441: t_0 $ and suppose that $ \theta_d^\prime(t_1) \not= 0. $ Because
442: \[E(\zeta,d) \geq E(\theta_d(t),\theta_d^\prime(t)) \geq \frac{1}{2}\theta_d^\prime(t_1)^2,\]
443: for all $ 0 \leq t\leq t_1. $ we deduce $ \frac{1}{2}d^2 >
444: \zeta(1-\frac{1}{3}\zeta^2), $ which contradicts (3.11). Therefore
445: $ \theta^\prime_d(t_1)= 0.$ In this case we deduce from the
446: equation of $ \theta_d $ that $ \theta_d^\sec(t_1) = 1 $ and then
447: $ \theta_d $ is nonnegative on a some neighbourhood of $ t_1.$
448: Consequently the local solution is nonnegative as long as there
449: exists. To show that $ \theta_d $ is global we note that
450: \[ E(\zeta,d) \geq \frac{1}{2}\theta_d^\prime(t)^2+\frac{1}{3}\theta_d^3(t)-\theta_d(t) \geq -\frac{2}{3},\]
451: for all $ t \leq T_d$, since $ \theta_d $ is nonnegative. Hence $
452: \theta_d $ and (then) $ \theta_d^\prime $ are bounded.
453: Consequently $ \theta_d $ is global.
454: It remains to show that $ \theta $ goes to unity at infinity.
455: To this end we use the {\it Bendixon Criterion}. Let ${\cal T}$
456: be the trajectory of $(\theta_d,\theta_d^\prime)$ in the phase
457: plane $ (0,\infty)\times\mathbb{R} $ for $ t \geq 0 $ and let
458: $\Gamma^+({\cal T})$ be its $w$-limit set at infinity. From the
459: boudedness of ${\cal T} $ it follows that $ \Gamma^+({\cal T})$
460: is a nonempty connected and compact subset of
461: $(0,\infty)\times\mathbb{R}$ ( see, for example \cite[p
462: 226]{Amann}). Moreover $ (-1,0) \not\in \Gamma^+({\cal T}),$ since
463: $ \theta_d $ is nonnegative. Note that if $\Gamma^+({\cal T})$
464: contains the equilibrium point $(1,0) $ then $\Gamma^+({\cal T}) =
465: \left\{(1,0)\right\},$ since $(1,0)$ is asymptotically stable.
466: Assume that $(1,0)\ \not\in\Gamma^+({\cal T}),$
467: Applying the {\it Poincar\'e--Bendixon Theorem }\cite[p 44]{Holmes} we deduce that $\Gamma^+({\cal T})$ is a
468: cycle, surrounding $(1,0)$. To finish, we shall
469: prove the nonexistence of such a cycle. We define $
470: P(\theta,\varphi) = \varphi, Q(\theta,\varphi) = -\t\varphi
471: +1-\theta^2, \varphi=\theta_d^\prime $ and $ \theta = \theta_d.$
472: The function $(\theta,\varphi)$ satisfies the system $ \theta^\prime = P(\theta,\varphi),\ \varphi^\prime = Q(\theta,\varphi).$ Let $ D $ be the bounded
473: domain of the $(\theta,\varphi)$--plane with boundary ${\Gamma^+}.
474: $ As $ P $ and $ Q $ are regular we deduce, via the {\it
475: Green--Riemann Theorem},
476: \begin{equation}
477: \int\int_D \left(\partial_\varphi Q + \partial_\theta
478: P\right)d\varphi d\theta=\int_{{\Gamma^+}}\left(Qd\theta - P
479: d\varphi\right)=0,
480: \end{equation}
481: thanks to the system satisfied by $(\theta,\varphi). $ But $
482: \partial_\varphi Q + \partial_\theta P = \t $ which is positive.
483: We get a contradiction.
484: \\
485: To complete our analysis, we shall determine a basin of the
486: critical point $(1,0).$ Let
487: \[{\cal P} = \left\{(\zeta,d)\ \in \mathbb{R}^2: \zeta > -1, \frac{1}{2}d^2 + \zeta\left(\frac{1}{3}\zeta^2-1\right) < \frac{2}{3} \right\}.\]
488: Let us consider a one-parameter of family of curves defined by
489: \[ E(\theta,\varphi) = \frac{1}{2}\varphi^2 + \frac{1}{3}\theta^3-\theta = C,\]
490: where $ C $ is a real parameter. Note that, in the phase plane,
491: this family is solution curves of system (3.4). The curve $
492: \varphi^2 = 2\theta - \frac{2}{3}\theta^3 + \frac{4}{3},$
493: corresponding to $ C = \frac{2}{3},$ goes through the point
494: $(2,0)$ and has the saddle $(-1,0) $ ($\t=0$) as its $ \alpha$ and
495: $w$-limit sets. We note this solution curve by $ {\cal H},$ which
496: is, in fact, an homoclinic orbit and define a separatrix cycle for
497: (3.4). We shall see that the bounded open domain with the boundary
498: $ {\cal H}$ is an attractor set for $(1,0)$ of system (3.9) where
499: $
500: \t > 0.$ This domain is given by $ E(\theta,\varphi) = C, \theta > -1, $ for all $ -\frac{2}{3} \leq C < \frac{2}{3}, $ which is $ {\cal P}.$ As $
501: \frac{d}{d t} E \leq 0 $ any solution, with initial data in $
502: {\cal P} $ cannot leave $ {\cal P}.$ By {\it LaSalle invariance
503: principle} we deduce that for any $(\zeta,d) $ in $ {\cal P } $
504: the $w$-limit set, $ \Gamma^+{(\zeta,d)} $ is a nonempty,
505: connected subset of $ {\cal P }\cap \left\{\varphi = 0\right\},$
506: (see \cite[p. 234]{Amann}). However, if $ \theta\not=1, \varphi = 0
507: $ is a transversal of the phase--flow, so the $w$--limit set is $
508: \left\{(1,0)\right\}.$ This means that $ {\cal P} $ is a basin
509: of the critical point $(1,0) $ of (3.9).\\
510: \begin{center}
511: \includegraphics[height=2.6 in]{basin.eps}
512: \begin{center}
513: { Fig. $3.2$} A basin of attraction ${\cal P }$ of the critical
514: point $(1,0)$.
515: \end{center}
516: \end{center}
517:
518: \section{Numerical results}
519: In this section numerical solutions of the boundary--value problem $(3.7)$ are obtained by using the fourth-order Runge-Kutta scheme with the shooting method. We plot the dimensionless velocity $\theta$ in term of the similarity variable $t$, for various value of the shooting parameter $d.$ \\
520:
521:
522: %---------------------
523: \bigskip
524: \begin{center}
525: \includegraphics[height=2.8in] {falkner2.eps}
526: \begin{center}
527: {Fig. $4.1$} Velocity profiles in terms of
528: $d=\theta^\prime(0)$ for fixed $\zeta =0.2$ and $\gamma =0.5$
529: \end{center}
530:
531: \end{center}
532: \bigskip
533: \begin{center}
534: \includegraphics[height=2.8in]{falkner3.eps}
535: \begin{center}
536: {Fig. $4.2$} Velocity profiles in terms of
537: $d=\theta^\prime(0)$ for fixed $\zeta =0.2$ and $\gamma =2.83$
538: \end{center}
539: \end{center}
540:
541:
542:
543:
544:
545: \section{Conclusion}
546: \setcounter{equation}{0} \setcounter{theorem}{0}
547: \setcounter{lemma}{0} \setcounter{remark}{0}
548: \setcounter{corollary}{0} \qquad In this work the laminar
549: two-dimensional steady incompressible, boundary layer flow past a
550: moving plane is considered. It has been shown that the problem
551: has solutions having a similarity form if the velocity
552: distribution outside the boundary layer is proportional to $ x^m,$
553: for some real number $ m.$ In the second part of this paper, we
554: are interested in question of existence of solutions in the case
555: where the external velocity is the inverse-linear function;
556: $m=-1.$ This situation occurs in the case of sink flow. To obtain
557: exact solutions the stream function $ \psi $ is written under the
558: form
559: \begin{equation}
560: \psi(x,y) = \sqrt{\nu u_\infty}f(t) + v_w\log(x).
561: \end{equation}
562: It is shown that
563: the
564: ordinary differential equation satisfied by $ f $ has multiple solutions for any $ v_w $ positive and no solution can exist if $ v_w
565: \leq 0.$ A sufficient condition for the existence is derived:
566: \begin{equation}\zeta > -1,\quad \quad \frac{1}{2}f^{\sec}(0)^2 + \zeta\left(\frac{\zeta^2}{3}-1\right) < \frac{2}{3}.\end{equation} We have
567: obtained two family of solutions according to $ \gamma =
568: v_w\left(\mu u_\infty\right)^{-1/2}. $ If $ \t \geq \sqrt{8} $,
569: $f^\prime $ is monotonic and goes to 1 at infinity, but if $ 0 <
570: \t < \sqrt{8},$ we have a stable spiral. The function $f^\prime $
571: oscillates an infinite number of times and goes to 1. So if we are
572: interested in solutions to $(3.7)$ such that
573: \[ -1 < f^\prime < 1 \] we must take $ u_w, v_w$ and $ u_\infty > 0 $ satisfying $ -u_\infty < u_w < u_\infty $ and $ v_w > (8\nu
574: u_\infty)^{1/2}.$\\
575: Condition (4.1) indicates also that for the same positive value of
576: the suction parameter the permeable wall stretching with velocity
577: $ u_wx^{-1}, u_w > 0 $ has multiple boundary--layer flows. Every
578: flow is uniquely determined by the dimensionless skin friction $
579: f^\sec(0) $ which can be any real number in the interval
580: $\left(-\sqrt{\frac{4}{3} +
581: 2\zeta\left(1-\zeta^2/3\right)},\sqrt{\frac{4}{3} +
582: 2\zeta\left(1-\zeta^2/3\right)}\right),$ where $ \zeta = u_w
583: u_\infty^{-1}.$ The case $ u_\infty = 0 $ was considered by
584: Magyari, Pop and Keller \cite{MPK}. The authors showed, by
585: numerical solutions, that the boundary
586: layer flow exists only for a large suction parameter ($\gamma \geq 1.079131$).\\
587: The existence of exact solutions of the Falkner-Skan equation under the
588: present condition was discussed by Rosenhead \cite[pp.
589: 244--246]{Rose} who mentioned that these results may be obtained
590: by rigorous arguments which, in fact, motivated the present work.
591: We note, in passing, that it is possible to obtain solutions if
592: the the skin friction satisfies $ \vert f^\sec(0)\vert >
593: \sqrt{\frac{4}{3} + 2\zeta\left(1-\zeta^2/3\right)}.$
594: \\
595:
596:
597:
598:
599: \section*{Appendix}
600: \qquad Let us now derive problem (3.5). We assume that the
601: external velocity is given by $ u_e(x) = u_\infty x^m,$ where $ m
602: $ is not necessary equal to -1. We recall that the stream function
603: satisfies the following equation Ä\begin{equation}
604: \partial_y\psi\partial_{xy}^2\psi - \partial_{x}\psi\partial_{yy}^2\psi =\nu\partial^3_{yyy}\psi+ mu_\infty^2x^{2m-1},
605: \end{equation}
606: with the boundary conditions
607: \begin{equation}
608: \partial_y\psi(x,0) = u_w x^{m},\quad \partial_x\psi(x,0) = -v_wx^{\frac{m-1}{2}},\quad \partial_y\psi(x,\infty)=u_\infty x^{m}.
609: \end{equation}
610: To obtain exact solutions to (5.3),(5.4), we look for
611: {\it ``pseudo-similarity " } solutions under the form
612: \begin{equation}\label{eq:Similarity}
613: \psi(x,y) = ax^{\alpha} F(x,byx^{-\beta}).
614: \end{equation}
615: where $\alpha =\frac{m+1}{2},\beta = -\frac{m-1}{2}, a =
616: \sqrt{\nu u_\infty} $ and $ b =\sqrt{ \frac{u_\infty}{\nu}}. $
617: Inserting (4.4) into (4.2),(4.3) leads to
618: \begin{equation}
619: \left\{\begin{array}{l} F^{\sec\prime} + \frac{1+m}{2} FF^{\sec}-
620: m({F^\prime}^2-1)+x\left(F^\sec\partial_xF-F^\prime\partial_xF^\prime\right)=0,\\
621: \\
622: F^\prime(x,0) = \zeta,\quad F^\prime(x,\infty)=1,\\
623: \\
624: \ds \frac{1+m}{2}F(x,0) +x\partial_x F(x,0) = \frac{v_w}{\sqrt{\nu
625: u_\infty}},
626: \end{array}\right.
627: \end{equation}
628: where the primes denote partial differential with respect to $ t =
629: \sqrt{\frac{u_\infty}{\nu}}yx^{\frac{m-1}{2}}.$ By writing
630: \[ F(x,t) = f(t) + H(x),\]
631: we find
632: \begin{equation}
633: \left\{\begin{array}{l}
634: f^{\sec\prime} + \frac{1+m}{2} ff^{\sec}- m\left({f^\prime}^2-1\right)+f^\sec\left(xH^\prime + \frac{1+m}{2}H\right)=0,\\
635: \\
636: \ds \frac{1+m}{2}f(0) +\frac{1+m}{2}H(x) + xH^\prime(x)=
637: \frac{v_w}{\sqrt{\nu u_\infty}},\quad f^\prime(0) = \zeta \in
638: [0,1),\quad f^{\prime}(\infty) = 1.
639: \end{array}\right.
640: \end{equation}
641: Hence, there exists a real number $ \t $ such that
642: \begin{equation}
643: \left\{\begin{array}{l}
644: f^{\sec\prime} + \frac{1+m}{2} ff^{\sec}- m\left({f^\prime}^2-1\right)+\t f^\sec=0,\quad t > 0,\\
645: \\
646: xH^\prime + \frac{1+m}{2}H = \t, \quad x > 0,\\
647: \\
648: \ds \frac{1+m}{2}f(0) +\t= \frac{v_w}{\sqrt{\nu u_\infty}},\quad
649: f^\prime(0) = \zeta,\quad f^{\prime}(\infty) = 1.
650: \end{array}\right.
651: \end{equation}
652: First, let us assume that $ m \not=-1.$ The solution $ H $ is given by
653: \[H(x) = cx^{-\frac{1+m}{2}}+ \frac{2\t}{1+m}, \]
654: where $ c $ is a constant, and then $ \psi(x,y) = ac +
655: ax^{\frac{1+m}{2}}\left(f(t) + \frac{2\t}{1+m}\right).$ The new
656: function $ g = f + \frac{2\t}{1+m} $ satisfies the Falkner-Skan
657: equation. Thereafter, we will assume that $ m = -1$ and this leads
658: to
659: \begin{equation}
660: \left\{\begin{array}{l}
661: f^{\sec\prime}+ \t f^\sec+ {f^\prime}^2-1=0,\\
662: \\
663: f^\prime(0) = \zeta,\quad f^{\prime}(\infty) = 1,\ \t=
664: \frac{v_w}{\sqrt{\nu u_\infty}},
665: \end{array}\right.
666: \end{equation}
667: and
668: \[ H(x) = \t\log x + c,\quad c = const.\]
669: \begin{quote} {\bf Acknowledgments. }{\small The authors would like to
670: thank R. Kersner for stimulating discussions and the referees
671: for their careful reading of the original manuscript and for
672: making constructive suggestions, which have improved the
673: presentation of this work. This work was partially supported by
674: Direction des Affaires Internationales (UPJV) Amiens, France and by PAI No MA/05/116. }
675: \end{quote}
676:
677: \begin{thebibliography}{99}
678:
679: \bibitem{Amann}
680: {\sc Amann H.,} Ordinary Differential Equations, An introduction
681: to nonlinear analysis, De Gruyter Studies in Mathematics, 13,
682: 1990.
683:
684: \bibitem{B7bis}
685: {\sc Banks W. H. H.,} { \it Similarity solutions of the boundary
686: layer equations for a stretching wall,} J. de M\'ecan. Th\'eo. et
687: Appl. { 2} (1983) 375--392.
688:
689: \bibitem{BZ4}
690: {\sc Banks W. H. H.} \& {\sc Zaturska M. B.,}{ \it
691: Eigensolutions in boundary--lary
692: flow adjacent to a stretching wall,} IMA Journal of Appl. Math. {
693: 36} (1986)
694: 375--392.
695:
696: \bibitem{Bar}
697: {\sc Barenblatt G. I.,} { Scaling, selfsimilar, and intermediate
698: asymptotics,} Cambridge Text in applied mathematics 14, 1996.
699: \bibitem{BBT3}
700: {\sc Belhachmi Z., Brighi B. \& Taous K.,} { \it On a family of
701: differential equation for boundary layer approximations in porous
702: media, } Euro. Jnl. Appl. Math. { 12} (2001) 513--528.
703: \bibitem{Blas}
704: {\sc Blasius H.,} {\it Grenzschichten in Fl\"{u}ssigkeiten mit
705: kleiner Reibung,} Z. math. Phys. 56 (1908) 1--37.
706: \bibitem{BC}
707: {\sc Bluman G. W. } and {\sc Cole J. D., } { \it Similarity
708: Methods for Differential Equations,} Applied Mathematical Sciences
709: 13 ( Springer, New York 1974).
710: \bibitem{CMP6}
711: {\sc Chaudhary M. A. Merkin J. H.} \& {\sc Pop, I.,} { \it
712: Similarity solutions in free convection boundary-layer flows
713: adjacent to vertical permeable surfaces in porous media. I.
714: Prescribed surface temperature,} Eur. J. Mech. B Fluids { 14} no.
715: 2 (1995) 217--237.
716: \bibitem{ChM1}
717: {\sc Cheng, P. \& Minkowycz, W.J.,} { \it Free convection about a
718: vertical flat plate embedded in a porous medium with application
719: to heat transfer from a dike,} J. Geophys. Res. { 82} (14)(1977)
720: 2040--2044.
721: \bibitem{CraPe}
722: {\sc Craven A.H. \& Peletier L. A.,} { \it On the uniqueness of
723: solutions of the Falkner--Skan Equation,} Mathematika {19} (1972)
724: pp. 129--133.
725: \bibitem{Cop11}
726: {\sc Coppel W. A.,} { \it On a differential equation of boundary
727: layer theory,} Phil. Trans. Roy. Soc. London, Ser. A { 253} (1960)
728: 101--136.
729:
730: \bibitem{E}
731: {\sc Evans H. L.,} { Laminar Boundary Layer Theory,}
732: Addison--Wesley, Reading, MA, 1968.
733:
734: \bibitem{FS}
735: {\sc Falkner v. M.} \& {\sc Skan S. W.,} { \it Solutions of
736: the boundary layer equations,} Phil. Mag., 12 (1931) 865--896.
737:
738: \bibitem{Holmes}
739: {\sc Guckenheimer J. } \& {\sc Holmes Ph.,} Nonlinear
740: oscillations, Dynamical systems, and Bifurcations of vector
741: fields, Applied Mathematical Sciences, 42, 1996.
742:
743: \bibitem{Guedda1}
744: {\sc Guedda M., } { \it Nonuniqueness of solutions to
745: differential equations for a boundary layer approximations in
746: porous media,} C. R. M\'ecanique, 300 (2002) 279--283.
747:
748: \bibitem{Guedda2}
749: {\sc Guedda M., } { \it Similarity solutions to differential
750: equations for boundary--layer approximations in porous media,} Z. angew. Math. Phys. (ZAMP), 56
751: (2005) 749--769.
752:
753:
754: \bibitem{HT1}
755: {\sc Hastings, S. P.} and {\sc Troy, W. C.,} {\it Oscillating
756: solutions of the Falkner-Skan equation for positive $\beta$,}
757: J. Diff. Eqn.,71 (1988) 123--144.
758: \bibitem{HT2}
759: {\sc Hastings, S. P.} and {\sc Troy, W. C.,}{\it Oscillating solutions of the Falkner-Skan equation for negative $\beta$.,} SIAM J. Math. Anal., 18
760: no. 2 (1987) 422--429.
761:
762: \bibitem{HS}
763: {\sc Hastings, S. P.} \& {\sc Siegel S.,} { \it On some solutions
764: of the Falkner--Skan Equation,} Mathematika {19} (1972)
765: 76--83.
766:
767: \bibitem{HJW}
768: {\sc Howell T. G., } {\sc JENG D. R. } \& {\sc De Witt K. J.,} {
769: \it Momentum and heat transfer on a continuous moving surface in a
770: power law fluid,} Int. J. Heat Mass Transfer, 40 (8) (1997)
771: 1853--1861.
772:
773: \bibitem{Ib1}
774: {\sc Ibragimov N. H.,} { \it CRC Handbook of Lie Groups Analysis
775: of Differential Eqiuations. I. Symmetries, Exacte Solutions and
776: Conservation laws } (CRC Press, Boca Raton 1993).
777:
778: \bibitem{Ib2}
779: {\sc Ibragimov N. H.,} { \it CRC Handbook of Lie Groups Analysis
780: of Differential Eqiuations. II. Applications in Engineering and
781: Physical Sciences } (CRC Press, Boca Raton 1994).
782:
783: \bibitem{JW}
784: {\sc Jones C. W.} \& {\sc Watson E.J.,} Two dimensional
785: boundary layers, in: Laminar Boundary Layer, L. Rosenhad ed.,
786: Dover Publ., New York, 1963, Chapt. V, sec 21.
787: \bibitem{LCB}
788: {\sc Ludlow D. K., Clarkson P. A.} and {\sc Bassom A. P., } {
789: \it New similarity solutions of the unsteady incompressible
790: boundary--layer equations,} Q. J. Mech. Appl. Math. 53 (2) (2000)
791: 175--206.
792: \bibitem{MK7}
793: {\sc Magyari E. } \& {\sc Keller B.,} { \it Exact solutions for
794: self-similar boundary-layer flows induced by permeable stretching
795: walls,} Eur. J. Mech. B Fluids { 19} no. 1 (2000) 109--122.
796: \bibitem{MPK}
797: {\sc Magyari E., Pop I. } \& {\sc Keller B.,} { \it The missing
798: similarity boundary--layer flow over a moving plane,} ZAMP {53}
799: (2002) 782--793.
800: \bibitem{Rose}
801: {\sc Rosenhead L.}, { Laminar Boundary layers,} Clarendon Press,
802: Oxford, 1963.
803: \bibitem{Sch}
804: {\sc Schlichting H.,} \& {\sc Gersten K.,} { Boundary layer
805: theory,} 8th revised and Enlarged Edition , McGraw Hill, New York,
806: 2000.
807: \bibitem{Sp}
808: {\sc Spalding D. B.,} { \it Mass transfer through laminar
809: boundary layers--1. The velocity boundary layer,} Int. J. Heat
810: Mass Transfer, Vol. 2 (1961) 15--32 .
811: \bibitem{Weyl}
812: {\sc Weyl H.,} { \it On the differential equations of the
813: simplest boundary--layer problems,} Ann. Math. { 253} (1942)
814: 381--407.
815:
816: \end{thebibliography}
817:
818:
819: \end{document}
820: