% vim: tw=50 % 02/03/2023 10AM \noindent eg Small oscillations of a gass bubble $a = a_0 + \eta(t)$, $|\eta| \ll a_0$. Linearise (1) $\implies$ $\rho a_0 \ddot{\eta} = p(a) - p_\infty$. Assume $p_\infty$ is constant. Gas in bubble obeys $pV^\gamma = \text{const}$ (** adiabatic variation in an ideal gas $\gamma = \frac{C_\rho}{C_V} = 1$ for air **) \myskip Then $\frac{\delta p}{p} = -\delta \frac{\delta V}{V} = -3\gamma \frac{\delta a}{a}$ because $V \propto a^3$. \[ \implies \rho a_0 \ddot{\eta} = p_0 (-3\gamma) \frac{\eta}{a_0} \] $\implies$ SHM with $\omega = \left( \frac{3\gamma p_0}{\rho a_0^2} \right)^{\half} \approx 2 \times 10^4 \mathsf{s^{-1}}$ for a $1\mathsf{mm}$ bubble. \newpage \section{Geophysical Flows} \subsection{Water Waves} Particularly successful application of potential flow. Consider \begin{center} \includegraphics[width=0.6\linewidth] {images/63627868b8e311ed.png} \end{center} \subsubsection{Governing Equations} Assume water is inviscid, and motion starts from rest. Hence, flow is and remains irrotational. \[ \implies \bf{u} = \nabla \phi \quad \text{and} \quad \boxed{\nabla^2 \phi = 0 \text{ in $-h < z < \zeta(x, y, t)$}} \tag{1} \] Kinematic boundary conditions \begin{itemize} \item At the rigid bottom \[ \boxed{\pfrac{\phi}{z} = 0 \text{ at } z = -h} \tag{2} \] \item At the air-water interface, often called the free surface as it is free to move, \[ \boxed{\pfrac{\zeta}{t} + u \pfrac{\zeta}{x} + v \pfrac{\zeta}{y} = w \text{ at } z = \zeta} \tag{3} \] (See section 1.4) \end{itemize} Dynamic boundary condition \[ \boxed{p_{\text{water}} = p_{\text{air}} \text{ at } z = \zeta(x, y, t)} \] As $p_{\text{air}} \ll p_{\text{water}}$, assume that pressure variations in the air are $\ll$ those in the water. Hence take $p_{\text{air}} = \text{const} = p_0$. Bernoulli for potential flow \[\rho \pfrac{\phi}{t} + \half \rho u^2 + p + \chi = f(t) \] independent of $\bf{x}$ ($\chi = \rho gz$). Apply to the surface and use $p = p_0 = \text{const}$. \[ \implies \boxed{\rho \pfrac{\phi}{t} + \half \rho |\nabla \phi|^2 + \rho g \zeta = f(t) \quad \text{ at } \quad z = \zeta(x, y, t)} \tag{4} \] Equations (1) - (4) give the full nonlinear problem. (1) is linear but on an unknown domain; (3) and (4) are complicated nonlinear boundary conditions to be applied at an unknown position $\zeta$. \subsubsection{Linear Water Waves} For small-amplitudes ($\zeta \ll h$ and $\pfrac{\zeta}{x}, \pfrac{\zeta}{y} \ll 1$, waveheight $\ll$ fluid depth and wavelenth) can linearise the prolem (about a state of rest): \begin{itemize} \item Ignore the terms quadratic in disturbance quantities, for example $u \pfrac{\zeta}{x}$, $\rho u^2$. \item Use Taylor series to expand boundary conditions at $z = \zeta$ in terms of information at $z = 0$. For example \[ \left. \pfrac{\phi}{z} \right|_{z = \zeta} = \left. \pfrac{\phi}{z} \right|_{z = 0} + \zeta \left. \pfrac[2]{\phi}{z} \right|_{z = 0} + \cdots \] (on the RHS we ignore everything other than the $\left. \pfrac{\phi}{z} \right|_{z = 0}$ term, as the others are quadratic or higher). \end{itemize} The linearise problem is \begin{align*} \nabla^2 &= 0 &&\text{in $-h < z < 0$} \tag{1} \\ \pfrac{\phi}{z} &= 0 &&\text{at $z = -h$} \tag{2} \\ \pfrac{\zeta}{t} &= \pfrac{\phi}{z} &&\text{at $z = 0$} \tag{3} \\ \rho \pfrac{\phi}{t} + \rho g \zeta &= f(t) &&\text{independent of $x, y$ at $z = 0$} \tag{4} \end{align*} Using IB Michaelmas Methods, we can \begin{itemize} \item either look for separable solutions $\phi(x, y, z, t) = \Phi(z) X(x) Y(y) T(t)$ and find \[ X'' = -k^2 X \quad Y'' = -l^2 T \quad T'' = -\omega^2 T \] or, on an infinite domain, take Fourier transforms with respect to $x, y, t$. \end{itemize} In 2D, end up looking for solution $\phi = e^{i(kx - \omega t)} \Phi(z)$, $\zeta = \zeta_0 e^{i(kx - \omega t)}$ (with real part understood). (1) $\nabla^2 \phi = 0$ implies $\Phi'' - k^2\Phi = 0$ ($*$). (Could write down $\Phi = e^{\pm kz}$, but bottom boundary condition $\Phi'(-h) = 0$ suggests cosh and sinh better with origin shifted to $z = -h$, ie:) \[ (*) \implies \phi = A \cosh(k (z + h)) + B \sinh(k(z + h)) \] and boundary condition (2) implies $B = 0$. In (4), LHS $\propto e^{ikx} \implies f(t) = 0$ and LHS $= 0$. So (3) and (4) \[ \begin{cases} -i\omega \zeta_0 = Ak\sinh kh \\ -i\omega A \cosh kh + g\zeta_0 = 0 \end{cases} \] Linear, homogeneous equations, with non-zero solution iff \[ \left| \begin{matrix} -k\sinh kh & -i\omega \\ -i\omega \cosh kh & g \end{matrix} \right| = 0 \] \[ \implies \omega^2 \cosh kh = gk \sinh kh \] Solutions must satisfy the \emph{dispersion relationship} \[ \boxed{\omega^2 = gk\tanh kh} \] Wave crest moves with the \emph{phase speed} \[ \boxed{c = \frac{\omega}{k}} \] Long waves propagate faster. \begin{center} \includegraphics[width=0.3\linewidth] {images/7fbf07ceb8e811ed.png} \end{center} Unlike light and sound, water waves are \emph{dispersion}: waves of different frequencies at different speeds and thus disperse as they propagate away from a local disturbance (splash, storm in the Atlantic).