Stokes’ 1st

Stokes' first problem, also erroneously called Rayleigh flow:

\begin{displaymath}
\hbox{\epsffile{figures/7-7.ps}}
\end{displaymath}

Continuity:

\begin{displaymath}
\mbox{div}\left(\vec v\right) = 0 =
\frac{\partial u}{\par...
...ac{\partial v}{\partial y}
\quad\Longrightarrow\quad u=u(y,t)
\end{displaymath}

$y$-momentum:

\begin{displaymath}
\rho \rlap{\kern 3pt\smash{\Bigg/}}\frac{Dv}{Dt} = - \rho g...
...rrow\quad p = -\rho g y + P(\rlap{\kern 0pt\smash{\bigg/}}x,t)
\end{displaymath}

$x$-momentum:

\begin{displaymath}
\rho \frac{\partial u}{\partial t}
+ \rho u \rlap{\kern 3p...
...u}{\partial x^2}
+ \frac{\partial^2 u}{\partial y^2}
\right)
\end{displaymath}

The $x$-momentum equation becomes:

\begin{displaymath}
u_t = \nu u_{yy}
\end{displaymath}

where $\nu=\mu/\rho$ is the kinematic viscosity.


\begin{displaymath}
\hbox{\epsffile{figures/7-7b.ps}}
\end{displaymath}

Exercise:

How would you normally find $u$?
$\bullet$

A simpler way to solve is to guess that the solution is similar: after rescaling $u$ and $y$, all velocity profiles look the same.

Original profiles:

\begin{displaymath}
\hbox{\epsffile{figures/7-7c.ps}}
\end{displaymath}

Supposed shape after scaling $u$ with $V_0$, and $y$ with a characteristic boundary layer thickness $\delta$ that increases with time:

\begin{displaymath}
\hbox{\epsffile{figures/7-7d.ps}}
\end{displaymath}

Mathematical form of the similarity assumption:

\begin{displaymath}
\frac{u}{V_0} = f\left(\frac{y}{\delta(t)},\rlap{\kern -3pt\smash{\bigg/}}t\right)
\end{displaymath}

The proof is in the pudding; if it satisfies the P.D.E., I.C., and B.C., it is OK.


\begin{displaymath}
u_t = \nu u_{yy} \quad\Longrightarrow\quad
-V_0 f' \frac{y}{\delta^2} \delta_t = \nu V_0 f'' \frac{1}{\delta^2}
\end{displaymath}

Put $\eta=y/\delta$:

\begin{displaymath}
-V_0 f' \eta \frac{\delta_t}{\delta} = \nu V_0 f'' \frac{1}{\delta^2}
\end{displaymath}

Separate into terms depending only on $\eta$ and terms depending only on $t$:

\begin{displaymath}
- \frac{f' \eta}{f''} = \frac{\nu}{\delta \delta_t} =
\mbox{constant} = \frac12
\end{displaymath}

It does not make a difference what you take the constant; this merely changes the value of $\delta$, not the physical solution.

Solving the O.D.E.s for $\delta$ and $f$, we solve the P.D.E. For the boundary layer thickness $\delta \delta_t = 2 \nu$ so

\begin{displaymath}
\delta = \sqrt{4\nu t}
\end{displaymath}

For the velocity profile $f''=-2\eta f'$ hence

\begin{displaymath}
f = \mbox{erfc}(\eta)
\end{displaymath}

where erfc is the complementary error function defined as

\begin{displaymath}
\fbox{$\displaystyle
\mbox{erfc}(x) \equiv \frac{2}{\sqrt{\pi}}\int_x^\infty e^{-\xi^2}\d \xi$}
\end{displaymath}

Exercise:

Derive the expressions for $\delta$ and $f$.
$\bullet$

Total:

\begin{displaymath}
\fbox{$\displaystyle
u = V_0 \mbox{ erfc}\left(\frac{y}{\delta}\right) \qquad
\delta = \sqrt{4\nu t} $}
\end{displaymath}

You should now be able to do 7.14, 16, 17