k-Wave User Forum » Topic: About laser ultrasound simulation

k-Wave User Forum » Topic: About laser ultrasound simulation http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation Support for the k-Wave MATLAB toolbox en-US Wed, 13 May 2026 06:52:38 +0000 http://bbpress.org/?v=1.0.2 <![CDATA[Search]]> q http://www.k-wave.org/forum/search.php Bradley Treeby on "About laser ultrasound simulation" http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation#post-7740 Wed, 19 Aug 2020 14:30:55 +0000 Bradley Treeby 7740@http://www.k-wave.org/forum/ Hi Alex, In this equation, I think the heat source term is in units of J.m^-3.s^-1. If you are adding the source to a fluid region of the domain, then you can just assign the same source term (in units of pascals) to the normal components of the stress. You will need to add a minus sign if you want the same polarity. Hope that helps, Brad. alexmm on "About laser ultrasound simulation" http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation#post-7694 Wed, 08 Jul 2020 22:33:46 +0000 alexmm 7694@http://www.k-wave.org/forum/ If I have a temperature field over time, would it be possible to add that as a stress source? alexmm on "About laser ultrasound simulation" http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation#post-7693 Wed, 08 Jul 2020 20:55:05 +0000 alexmm 7693@http://www.k-wave.org/forum/ Hi Brad, Thank you for your answer. I am now trying to use the same model and include a 2d model for the laser source. I have an equation that can deposit a given power per unit volume at a given location and depth and time. In the k-wave manual pg 8 it says I can add a mass source using S_M = S_H*beta*C_p. For that equation, what are the units of S_H? Are they J/m^3? What if I have a changing power over time? For the pstdElastic2D solver I can only add stress sources that change over time. Is there a mesh dependent conversion factor I need to use to change S_M to a stress source? I'm guessing this same value has to be added to xx and yy components is this correct? Thank you, Alex Bradley Treeby on "About laser ultrasound simulation" http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation#post-7653 Sat, 27 Jun 2020 13:20:00 +0000 Bradley Treeby 7653@http://www.k-wave.org/forum/ Hi Alex, Regarding simulating air boundaries, spectral methods sometimes struggle as discussed in the <a href="http://www.k-wave.org/forum/topic/simulation-of-ultrasonic-testing-borehole-in-solid">this post</a>. By default, the outer boundaries are always absorbing (a perfectly matched layer is used). In principle, it's straightforward to impose Dirichlet and Neumann boundary conditions on the grid boundary (as discussed in <a href="https://arxiv.org/pdf/2005.00322">this paper</a> with code examples <a href="https://www.mathworks.com/matlabcentral/fileexchange/75071-matlab-discrete-trigonometric-transform-library">here</a>). However, this doesn't currently form part of k-Wave. You will need the parallel computing toolbox and a compatible GPU to use the <code>'DataCast'</code> GPU options. What is the error you receive? Can you run native MATLAB code on the GPU (try casting something to <code>gpuArray</code>). I don't know a reference off-hand for this unfortunately, but it's likely some one has tried it before. Brad. alexmm on "About laser ultrasound simulation" http://www.k-wave.org/forum/topic/about-laser-ultrasound-simulation#post-7631 Fri, 19 Jun 2020 12:36:53 +0000 alexmm 7631@http://www.k-wave.org/forum/ Hi, I’d like to say I’ve found the k-wave software really useful. Because of COVID I couldn’t access lab equipment to run an experiment and I’m using k-wave to simulate it. It’s really good work. What I’m using now is the elastic solver in 2d to simulate surface motion from a laser ultrasound setup for NDT. The code is the following: <pre><code>close all clear all %Physical parameters (all in SI units) %Air properties factor = 2.0; rho_air = 1.225/factor^2; c_long_air = 343.0*factor; c_shear_air = 0.0; %Aluminum properties (pg 273 Scruby book) rho_Al = 2700.0; c_long_Al = 6400.0; c_shear_Al = 3150.0; %Laser properties t_pulse_width = 10.0e-9; %(10 ns) f_max = 0.1874/t_pulse_width %Simulation domain x_Al = 2.00e-3; x_air = 0.56e-3; y_size_length = 7.68e-3; %Factor of 3 instead of 2 %Generation laser pulse width %Compute dx and Nx based on a desired x_size and f_max points_per_wavelength = 3; %Can use min of 3, better to use higher number max_possible_dx = min([c_long_air c_long_Al c_shear_Al])/f_max/points_per_wavelength dx = 5.0e-6*factor %Try to make these numbers a power of 2 (pg 30 of k-Wave manual) Nx_Al = round(x_Al/dx) Nx_air = round(x_air/dx) Nx = Nx_Al+Nx_air Ny = round(y_size_length/dx) %Create computational grid kgrid = kWaveGrid(Nx,dx, Ny,dx); %Define the compressional sound speed [m/s] medium.sound_speed_compression = c_long_Al*ones(Nx, Ny); medium.sound_speed_compression((Nx_Al+1):end,:) = c_long_air; %Define the shear sound speed [m/s] medium.sound_speed_shear = c_shear_Al*ones(Nx, Ny); medium.sound_speed_shear((Nx_Al+1):end,:) = c_shear_air; %Define the mass density [kg/mˆ3] medium.density = rho_Al*ones(Nx, Ny); medium.density((Nx_Al+1):end,:) = rho_air; %Define the absorption coefficients [dB/(MHzˆ2 cm)] medium.alpha_coeff_compression = 0.0; medium.alpha_coeff_shear = 0.0; %Define circular through hole (along width) depth = 1.0e-3; diam = 0.5e-3; y_position = round(0.5e-3/dx); x_position = round((x_Al-depth)/dx); radius = round(diam/2.0/dx); plot_disc = 0; disc = makeDisc(Nx,Ny, x_position,y_position, radius, plot_disc); medium.sound_speed_compression(disc == 1) = c_long_air; medium.sound_speed_shear(disc == 1) = c_shear_air; medium.density(disc == 1) = rho_air; figure; surf(medium.density) %Create the time array cfl = 0.008; %(0.02 with air props) % Courant-Friedrichs-Lewy number t_end = 1.2e-6; % [s] kgrid.makeTime(max(medium.sound_speed_compression(:)), cfl, t_end); %Define an initial pressure distribution using makeBall height_from_bottom = 1.9e-3; from_side = 5.0e-3; source.p0 = 2.0*makeDisc(Nx,Ny,round(height_from_bottom/dx),round(from_side/dx), 5); %Radius should be 10 %Define a 3D binary sensor mask for a single grid point from_side = 2.0e-3; sensor.mask = zeros(Nx,Ny); sensor.mask(Nx_Al,round(from_side/dx)) = 1; sensor.record = {'p','u'}; %Plot sound source and sensor locations figure; surf(logical(source.p0)+sensor.mask) %Run the simulation input_args = {'PlotScale',[-0.75, 0.75, -0.15, 0.15], 'PlotPML',false, 'DisplayMask',zeros(Nx,Ny), 'DataCast','single', ... 'RecordMovie',true, 'MovieName','example_movie', 'PlotScale',[-0.1, 0.1], 'PlotFreq',50, 'MovieProfile','MPEG-4'}; sensor_data = pstdElastic2D(kgrid, medium, source,sensor, input_args{:}); %Plot A-scan data figure; plot(kgrid.t_array, sensor_data.ux)</code></pre> I have a few questions regarding a more accurate simulation and computation speed. • Right now I’m simulating an Aluminum sample with a through hole and a free surface (with air properties). Instead of having an air layer, how could I enforce a free layer (reflecting?) boundary condition instead of a sample air interface. Are the BCs absorbing by default? • I tried using the gpu option but it’s not working at the moment. Do I have to install something else to make this work? • Is there an example (papers and/or code) of using a point, line (or half circle?) with a certain directivity function and time profile to more accurately simulate a laser pulse? Thank you, Alex