notes

uni notes
git clone git://popovic.xyz/notes.git
Log | Files | Refs

chap2.tex (9665B)


      1 \section{Dimensional Analysis}
      2 Our derived model of fluid mechanics yields formal connections between
      3 physical quantities. These quantities bear units, e.g. the velocity of fluid
      4 particles $\mathbf{u}$ has the ``SI'' unites of $\frac{m}{s}$, meters per
      5 second. The idea is to make use of these scales and formulate a model, where
      6 the quantities are nondimensionalized, i.e. to get rid of physical units by
      7 scaling each quantity appropriately. The appropriate length scales are that
      8 of the typical water depth $h_0$ and the typical wavelength $\lambda$ of a
      9 surface wave.
     10 
     11 \subsection{Nondimensionalisation\label{sec:nondim}}
     12 In summary we use these adaptations for the scales
     13 
     14 \begin{itemize}
     15     \item $h_0$ for the typical water depth
     16     \item $\lambda$ for the typical wavelength
     17     \item $\frac{\lambda}{\sqrt{g h_0}}$ time scale of wave propagation
     18     \item $\sqrt{g h_0}$ velocity scale of waves in $(x, y)$
     19     \item $\frac{h_0 \sqrt{g h_0} }{\lambda}$ velocity scale in the $z$
     20         direction.
     21 \end{itemize}
     22 Additionally we write the
     23 boundary condition on the free surface as follows
     24 \begin{align}
     25     h  = h_0 + a \eta (\mathbf{x}_\perp, t) = z,
     26 \end{align}
     27 where $a$ is the typical amplitude and $\eta$ nondimensional function. All in
     28 all we have the following scaling for the physical quantities of our context
     29 \begin{align}
     30     &x \rightarrow\ \lambda x, \quad u \rightarrow \sqrt{gh_0} u, \\
     31       &y \rightarrow\ \lambda y, \quad v \rightarrow \sqrt{gh_0} v, \qquad
     32       t\rightarrow \frac{\lambda}{\sqrt{gh_0}}t,\\
     33       &z \rightarrow\ h_0 z, \quad w \rightarrow
     34     \frac{h_0\sqrt{gh_0}}{\lambda} w.
     35 \end{align}
     36 with
     37 \begin{align}
     38     h = h_0 + a \eta, \qquad  b \rightarrow h_0 b.
     39 \end{align}
     40 The pressure is also rewritten into
     41 \begin{align}
     42     P = P_a + \rho g(h_0 -z) + \rho g h_0 p,
     43 \end{align}
     44 where $P_a$ is the atmospheric pressure, the term $h_0-z$ represent the
     45 hydrostatic pressure distribution, i.e. pressure at depth and the term with the pressure
     46 variable $p$  measures the deviation from the hydrostatic pressure
     47 distribution. We have $p\neq 0 $ for wave propagation. Now we can perform a
     48 rescaling of the Euler's Equation of Motion, we introduce the notation
     49 \begin{align}
     50     &t = \frac{\lambda}{\sqrt{gh_0}}\tau,\quad x = \lambda \xi,\quad u =
     51     \sqrt{gh_0} \tilde{u}\\
     52     &y = \lambda \chi,\quad v = \sqrt{gh_0} \tilde{v}\\
     53     &z = h_0 \zeta, \quad w = \frac{h_0\sqrt{gh_0} }{\lambda}\tilde{w}.
     54 \end{align}
     55 We start off with the $x$ coordinate, substitute and apply the chain rule
     56 leaving us with
     57 \begin{align}
     58     \frac{Du}{Dt}
     59     &= \frac{\partial u}{\partial t} +u \frac{\partial
     60     u}{\partial x} \\
     61     &= \sqrt{gh_{0}}\frac{\partial \tilde{u}}{\partial \tau} \frac{\partial
     62     \tau}{\partial t} +gh_0 \tilde{u} \frac{\partial \tilde{u}}{\partial \xi}
     63     \frac{\partial \xi}{\partial x} \\
     64     &= \frac{gh_0}{\lambda} \left( \frac{\partial \tilde{u}}{\partial \tau}
     65     + \tilde{u} \frac{\partial \tilde{u}}{\partial \xi} \right),
     66 \end{align}
     67 on the other hand
     68 \begin{align}
     69     \frac{gh_0}{\lambda} \left( \frac{\partial \tilde{u}}{\partial \tau}
     70     +\tilde{u} \frac{\partial \tilde{u}}{\partial \xi} \right)
     71     &=-\frac{1}{\rho}\frac{1}{\lambda}\frac{\partial P}{\partial x} \\
     72     &=-\frac{ g h_0 }{\lambda}\rho \frac{\partial p}{\partial \xi}.
     73 \end{align}
     74 Thereby the rescaling evolves to
     75 \begin{align}
     76     \frac{D \tilde{u}}{D\tau} = -\frac{\partial p}{\partial \xi}.
     77 \end{align}
     78 Because of the same scaling in $y$ we get the same result as in $x$, that is
     79 \begin{align}
     80     \frac{D \tilde{v}}{D\tau} = -\frac{\partial p}{\partial \chi}.
     81 \end{align}
     82 In the $z$ coordinate we have
     83 \begin{align}
     84     \frac{Dw}{Dt}
     85     &= \frac{\partial w}{\partial t} +w \frac{\partial
     86     w}{\partial \zeta} \\
     87     &= \frac{h_0\sqrt{gh_0}}{\lambda} \frac{\sqrt{gh_0}}{\lambda}
     88     \frac{\partial \tilde{w}}{\partial \tau}  + \frac{1}{h_0}
     89     \frac{h_0\sqrt{gh_0} }{\lambda} \frac{h_0\sqrt{gh_0}}{\lambda}
     90     \tilde{w}\frac{\partial \tilde{v}}{\partial \zeta}\\
     91     &= \frac{h_0^2g}{\lambda}\left( \frac{\partial \tilde{w}}{\partial \tau}
     92     + \tilde{w}\frac{\partial \tilde{w}}{\partial \zeta} \right) .
     93 \end{align}
     94 On the other side we have
     95 \begin{align}
     96     \frac{h_0^2g}{\lambda^2}\left( \frac{\partial \tilde{w}}{\partial \tau}
     97     + \tilde{w}\frac{\partial \tilde{w}}{\partial \zeta} \right)
     98     &=
     99     -\frac{1}{h_0\rho} \frac{\partial P}{\partial z} +g \\
    100     &=-\frac{1}{h_0\rho}\left(-\rho gh_0 \frac{\partial \zeta}{\partial
    101         \zeta} + \rho gh_0
    102     \frac{\partial p}{\partial \zeta} \right) + g  \\
    103     &= -g \frac{\partial p}{\partial \zeta}.
    104 \end{align}
    105 In total for the $z$ direction we get
    106 \begin{align}
    107    \underbrace{\left( \frac{h_0}{\lambda} \right)^2}_{=: \delta^2}
    108     \frac{Dw}{Dt} = -\frac{\partial p}{\partial \zeta},
    109 \end{align}
    110 where $\delta$ is the \textbf{long wavelength} or \textbf{shallowness}
    111 parameter, a very important constant for developing model hierarchies. For
    112 clarity we resubstitute for $x, y, z, t, u, v$ and $w$, and for completeness
    113 we display the equations again, which are
    114 \begin{align}\label{eq:nondim-motion}
    115     \frac{Du}{Dt} = - \frac{\partial p}{\partial x}&, \quad
    116     \frac{Dv}{Dt} = - \frac{\partial p}{\partial y}, \quad
    117     \delta^2\frac{Dw}{Dt} = - \frac{\partial p}{\partial z}, \\
    118     &\frac{\partial u}{\partial x} + \frac{\partial v}{\partial y}
    119     +\frac{\partial w}{\partial z}  = 0.
    120 \end{align}
    121 We can now turn our attention to the boundary conditions, on both free
    122 surface $z=h$ and the bottom $z=b$ we have $z \Rightarrow h_0 z$ and thereby
    123 \begin{align}
    124     z = 1+
    125     \underbrace{\frac{a}{h_0}}_{:=\varepsilon}\eta(\mathbf{x}_\perp,t) \quad
    126     \text{and}\quad z= b,
    127 \end{align}
    128 where we arrive at our second very important parameter $\varepsilon$ called
    129 the \textbf{amplitude} parameter. As for the kinematic condition, we
    130 substitute the free surface $z=h = 1+\varepsilon \eta$ and get
    131 \begin{align}
    132     \frac{Dz}{Dt} = \varepsilon\left(\eta_t + (\mathbf{u}_\perp
    133         \nabla_\perp)\eta\right) \qquad \text{on}\;\; z= 1+\varepsilon \eta.
    134 \end{align}
    135 Respectively the bottom condition is not changed
    136 \begin{align}
    137     w = b_t + (\mathbf{u}_\perp \nabla_\perp) b \quad \text{on}\;\; z= b.
    138 \end{align}
    139 The general dynamic condition for $h = h(x, y, t)$ yields a rescaling of the
    140 curvature in terms of
    141 \begin{align}
    142    \frac{1}{R}
    143    &= \frac{(1+h_y^2)h_{x x} + (1+h_x^2)h_yy - 2h_xh_yh_{xy}
    144    }{\left(h_x^2+h_y^2 +1  \right)^{\frac{3}{2}} } \\
    145    &= -\frac{\varepsilon h_0}{\lambda^2} \frac{(
    146    1+\varepsilon^2\delta^2\eta_y^2 )\eta_{x x}+
    147     (1+\varepsilon^2\delta^2\eta_x^2)\eta_{yy} -
    148     2\varepsilon^2\delta^2\eta_x\eta_y\eta_{xy}}{\left(
    149     1+\varepsilon^2\delta^2\eta_x^2+\varepsilon^2\delta^2\eta_y^2
    150     \right)^{\frac{3}{2}} },
    151 \end{align}
    152 together with the pressure difference
    153 \begin{align}
    154     \Delta P = \rho g h_0(p - \varepsilon \eta) = \frac{\Gamma}{R},
    155 \end{align}
    156 leaving us ultimately with the dynamic condition
    157 \begin{align}
    158     p-\varepsilon\eta= \varepsilon\left( \frac{\Gamma}{\rho g\lambda^2}
    159     \right) \left(\frac{\lambda^2}{\varepsilon h_0}\frac{1}{R}\right),
    160 \end{align}
    161 where $W_e = \frac{\Gamma}{\rho g h_0^2}$ is the \textbf{Weber number}. This
    162 dimensionless parameter can be considered as a measure of the fluid's inertia
    163 compered to its surface tension, which satisfies the relation
    164 \begin{align}
    165     \delta^2 W_e = \frac{\Gamma}{\rho g \lambda^2}.
    166 \end{align}
    167 \subsection{Scaling of Variables}
    168 Admits a simple observation of the governing equations in the last chapter we
    169 notice that $w$ and $p$ on the free surface $z = 1 + \varepsilon\eta$ are
    170 directly proportional to $\varepsilon$. Hence we want to ''scale this way``
    171 by introducing the following transformation
    172 \begin{align}
    173     p \rightarrow \varepsilon p, \quad w \rightarrow \varepsilon w, \quad
    174     \mathbf{u}_\perp \rightarrow \varepsilon \mathbf{u}_\perp.
    175 \end{align}
    176 Because of this scaling our material derivative changes slightly to
    177 \begin{align}\label{eq:mod-material}
    178     \frac{D}{Dt} = \frac{\partial }{\partial t} + \varepsilon\left(u
    179     \frac{\partial }{\partial x}  + v \frac{\partial }{\partial y}  + w
    180     \frac{\partial }{\partial z} \right)
    181 \end{align}
    182 A simple recalculation yields the rescaled, nondimensionalized Euler's
    183 Equation of motion are the same as in equations \ref{eq:nondim-motion} with
    184 the modified material derivative from \ref{eq:mod-material}, and the boundary
    185 conditions are
    186 \begin{align}
    187     \begin{drcases}
    188     p = \eta - \frac{\delta^2\varepsilon h_0}{\lambda^2} \frac{W_e}{R}\\
    189     w = \frac{1}{\varepsilon}\eta_t + (\mathbf{u}_\perp \nabla_\perp)\eta
    190     \end{drcases} \quad
    191     \text{on}\;\; z = 1+\varepsilon\eta\\
    192     w =\frac{1}{\varepsilon}b_t + (\mathbf{u}_\perp \nabla_\perp)b \quad
    193     \text{on}\;\; z=b
    194 \end{align}
    195 \subsection{Model Hierarchies}
    196 As we have derived a model of fluid mechanics, with small parameters
    197 $\varepsilon$ and $\delta$, we can conduct a series of classifications on the
    198 long wave problem and perform asymptotic analysis on them. The main
    199 hierarchies important in this review are derived from the following problem
    200 classifications
    201 \begin{itemize}
    202     \item $\varepsilon\rightarrow 0$: linearized problem, small amplitude
    203     \item $\delta\rightarrow 0$: shallow Water, long-wave
    204     \item$\delta \rightarrow 0;\; \varepsilon\approx1$: shallow Water, large
    205         amplitude
    206     \item $\delta\ll 1;\; \varepsilon\approx\delta$: shallow water, medium
    207         amplitude
    208     \item $\delta\ll 1;\; \varepsilon\approx\delta^2$: shallow water, small
    209         amplitude
    210     \item $\delta \gg 1;\; \varepsilon\cdot\delta\ll 1$: deep water, small
    211         steepness.
    212 \end{itemize}
    213 
    214 
    215