The source free elastic wave equation can be written in terms of a coupled first-order system
\begin{align} \partial_t \sigma - \mu \partial_x v & = 0 \\ \partial_t v - \frac{1}{\rho} \partial_x \sigma & = 0 \end{align}with $\rho$ the density and $\mu$ the shear modulus. This equation in matrix-vector notation follows
\begin{equation} \partial_t \mathbf{Q} + \mathbf{A} \partial_x \mathbf{Q} = 0 \end{equation}where $\mathbf{Q} = (\sigma, v)$ is the vector of unknowns and the matrix $\mathbf{A}$ contains the parameters $\rho$ and $\mu$. The above matrix equation is analogous to the advection equation $\partial_t q + a \partial_x q = 0$. Although it is a coupled system, diagonalization of $\mathbf{A} = \mathbf{R}^{-1}\mathbf{\Lambda}\mathbf{R}$ allows us to implement all elements developed for the solution of the advection equation in terms of fluxes. It turns out that the decoupled version is
\begin{equation} \partial_t \mathbf{W} + \mathbf{\Lambda} \partial_x \mathbf{W} = 0 \end{equation}where the eigenvector matrix $\mathbf{R}$ and the diagonal matrix of eigenvalues $\mathbf{\Lambda}$ are given
\begin{equation} \mathbf{W} = \mathbf{R}^{-1}\mathbf{Q} \qquad\text{,}\qquad \mathbf{\Lambda}= \begin{pmatrix} -c & 0 \\ 0 & c \end{pmatrix} \qquad\text{,}\qquad \mathbf{R} = \begin{pmatrix} Z & -Z \\ 1 & 1 \end{pmatrix} \qquad\text{and}\qquad \mathbf{R}^{-1} = \frac{1}{2Z} \begin{pmatrix} 1 & Z \\ -1 & Z \end{pmatrix} \end{equation}Here $Z = \rho c$ with $c = \sqrt{\mu/\rho}$ represents the seismic impedance.
This notebook implements both upwind and Lax-Wendroff schemes for solving the free source version of the elastic wave equation in a homogeneous media. To keep the problem simple we use as spatial initial condition a Gauss function with half-width $\sigma$
\begin{equation} Q(x,t=0) = e^{-1/\sigma^2 (x - x_{o})^2} \end{equation}
In [ ]:
# Import all necessary libraries, this is a configuration step for the exercise.
# Please run it before the simulation code!
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.animation as animation
# Show the plots in the Notebook.
plt.switch_backend("nbagg")
In [ ]:
# Initialization of setup
# --------------------------------------------------------------------------
nx = 800 # number of grid points
c = 2500 # acoustic velocity in m/s
ro = 2500 # density in kg/m^3
Z = ro*c # impedance
mu = ro*c**2 # shear modulus
xmax = 10000 # Length in m
eps = 0.5 # CFL
tmax = 2.0 # simulation time in s
isnap = 10 # plotting rate
sig = 200 # argument in the inital condition
x0 = 5000 # position of the initial condition
imethod = 'upwind' # 'Lax-Wendroff', 'upwind'
# Initialize Space
x, dx = np.linspace(0,xmax,nx,retstep=True)
# use wave based CFL criterion
dt = eps*dx/c # calculate time step from stability criterion
# Simulation time
nt = int(np.floor(tmax/dt))
# Initialize wave fields
Q = np.zeros((2,nx))
Qnew = np.zeros((2,nx))
Qa = np.zeros((2,nx))
Seismic disturbances are introduced through specification of a particular spatial initial condition, in this case we use a Gaussian function.
Implement the spatial initial condition given by:
\begin{equation} Q(x,t=0) = e^{-1/\sigma^2 (x - x_{o})^2} \end{equation}Then, visualize the initial condition in a given plot.
In [1]:
#################################################################
# INITIALIZE THE SOURCE TIME FUCTION HERE!
#################################################################
#################################################################
# PLOT THE SOURCE TIME FUNCTION HERE!
#################################################################
We decompose the solution into right propagating $\mathbf{\Lambda}^{+}$ and left propagating eigenvalues $\mathbf{\Lambda}^{-}$ where
\begin{equation} \mathbf{\Lambda}^{+}= \begin{pmatrix} -c & 0 \\ 0 & 0 \end{pmatrix} \qquad\text{,}\qquad \mathbf{\Lambda}^{-}= \begin{pmatrix} 0 & 0 \\ 0 & c \end{pmatrix} \qquad\text{and}\qquad \mathbf{A}^{\pm} = \mathbf{R}^{-1}\mathbf{\Lambda}^{\pm}\mathbf{R} \end{equation}This strategy allows us to formulate an upwind finite volume scheme for any hyperbolic system as
\begin{equation} \mathbf{Q}_{i}^{n+1} = \mathbf{Q}_{i}^{n} - \frac{dt}{dx}(\mathbf{A}^{+}\Delta\mathbf{Q}_{l} - \mathbf{A}^{-}\Delta\mathbf{Q}_{r}) \end{equation}with corresponding flux term given by
\begin{equation} \mathbf{F}_{l} = \mathbf{A}^{+}\Delta\mathbf{Q}_{l} \qquad\text{,}\qquad \mathbf{F}_{r} = \mathbf{A}^{-}\Delta\mathbf{Q}_{r} \end{equation}The upwind solution presents a strong diffusive behaviour. In this sense, the Lax-Wendroff perform better, with the advantage that it is not needed to decompose the eigenvalues into right and left propagations. Here the matrix $\mathbf{A}$ can be used in its original form. The Lax-Wendroff follows
\begin{equation} \mathbf{Q}_{i}^{n+1} = \mathbf{Q}_{i}^{n} - \frac{dt}{2dx}\mathbf{A}(\mathbf{Q}_{i+1}^{n} - \mathbf{Q}_{i-1}^{n}) + \frac{1}{2}\Big(\frac{dt}{dx}\Big)^2\mathbf{A}^2(\mathbf{Q}_{i+1}^{n} - 2\mathbf{Q}_{i}^{n} + \mathbf{Q}_{i-1}^{n}) \end{equation}Initialize all relevant matrices, i.e $R$, $R^{-1}$, $\mathbf{\Lambda}^{+}$, $\mathbf{\Lambda}^{-}$, $\mathbf{A}^{+}$, $\mathbf{A}^{-}$, $\mathbf{A}$.
In [2]:
#################################################################
# INITIALIZE ALL MATRICES HERE!
#################################################################
# R =
# Rinv =
# Lp =
# Lm =
# Ap =
# Am =
# A =
In [ ]:
# Initialize animated plot
# ---------------------------------------------------------------
fig = plt.figure(figsize=(10,6))
ax1 = fig.add_subplot(2,1,1)
ax2 = fig.add_subplot(2,1,2)
line1 = ax1.plot(x, Q[0,:], 'k', x, Qa[0,:], 'r--')
line2 = ax2.plot(x, Q[1,:], 'k', x, Qa[1,:], 'r--')
ax1.set_ylabel('Stress')
ax2.set_ylabel('Velocity')
ax2.set_xlabel(' x ')
plt.suptitle('Homogeneous F. volume - %s method'%imethod, size=16)
plt.ion() # set interective mode
plt.show()
# ---------------------------------------------------------------
# Time extrapolation
# ---------------------------------------------------------------
for i in range(nt):
if imethod =='Lax-Wendroff':
for j in range(1,nx-1):
dQ1 = Q[:,j+1] - Q[:,j-1]
dQ2 = Q[:,j-1] - 2*Q[:,j] + Q[:,j+1]
Qnew[:,j] = Q[:,j] - 0.5*dt/dx*(A @ dQ1)\
+ 1./2.*(dt/dx)**2 * (A @ A) @ dQ2 # Eq. 8.56
# Absorbing boundary conditions
Qnew[:,0] = Qnew[:,1]
Qnew[:,nx-1] = Qnew[:,nx-2]
elif imethod == 'upwind':
for j in range(1,nx-1):
dQl = Q[:,j] - Q[:,j-1]
dQr = Q[:,j+1] - Q[:,j]
Qnew[:,j] = Q[:,j] - dt/dx * (Ap @ dQl + Am @ dQr) # Eq. 8.54
# Absorbing boundary conditions
Qnew[:,0] = Qnew[:,1]
Qnew[:,nx-1] = Qnew[:,nx-2]
else:
raise NotImplementedError
Q, Qnew = Qnew, Q
# --------------------------------------
# Animation plot. Display solution
if not i % isnap:
for l in line1:
l.remove()
del l
for l in line2:
l.remove()
del l
# --------------------------------------
# Analytical solution (stress i.c.)
Qa[0,:] = 1./2.*(np.exp(-1./sig**2 * (x-x0 + c*i*dt)**2)\
+ np.exp(-1./sig**2 * (x-x0-c*i*dt)**2))
Qa[1,:] = 1/(2*Z)*(np.exp(-1./sig**2 * (x-x0+c*i*dt)**2)\
- np.exp(-1./sig**2 * (x-x0-c*i*dt)**2))
# --------------------------------------
# Display lines
line1 = ax1.plot(x, Q[0,:], 'k', x, Qa[0,:], 'r--', lw=1.5)
line2 = ax2.plot(x, Q[1,:], 'k', x, Qa[1,:], 'r--', lw=1.5)
plt.legend(iter(line2), ('F. Volume', 'Analytic'))
plt.gcf().canvas.draw()
In [ ]: