1887
banner image
No data available.
Please log in to see this content.
You have no subscription access to this content.
No metrics data to plot.
The attempt to load metrics for this article has failed.
The attempt to plot a graph for these metrics has failed.
Fourier continuation methods for high-fidelity simulation of nonlinear acoustic beams
Rent:
Rent this article for
USD
10.1121/1.4742722
/content/asa/journal/jasa/132/4/10.1121/1.4742722
http://aip.metastore.ingenta.com/content/asa/journal/jasa/132/4/10.1121/1.4742722
View: Figures

Figures

Image of FIG. 1.
FIG. 1.

An illustration of the main elements of the FC method: after producing a translated copy of the given values of f on the interval [b, 1 + b], the algorithm proceeds to (1) obtain a smooth function that blends the original function on the interval [0, 1] to zero in the interval [1, b], and, similarly, to obtain a smooth function that blends the original (translated) function on the interval [b, 1 + b] to zero in the interval [1, b]; and to (2) add these two blending functions to obtain a smooth continuation. The original function values are represented by solid circles and the additional continuation values are represented by open circles. The thick curves denote the polynomial approximations in the matching intervals. The thin gray lines show the blending functions that are summed to produce the continuation values. The Fourier continuation fc then results by a direct application of the FFT algorithm.

Image of FIG. 2.
FIG. 2.

A simplified illustration of the spatial discretization used by the FC-based solver for the advection Eq. (6). (a) N = 30 grid points are evenly divided into M = 3 subdomains. (b) The curve represents a function that is to be differentiated by means of the FC approach. The derivative values at the points in the right-most subdomain [black circles in (a)] are computed using the function values at these points as well as external fringe region points [gray circles in (a)]; the function values on this augmented set ofpoints is represented by the thicker portion of the curve in (b). This (non-periodic) function is differentiated by means of the FC differentiation algorithm. The derivative values in the remaining two subdomains are produced analogously.

Image of FIG. 3.
FIG. 3.

(Top left) Initial conditions for the velocity u in the one-dimensional advection equation example presented in Sec. IV A. (Top right) Close-up view of the initial condition near x = 0. (Bottom left and right) Errors vs computing time and discretization size, respectively at time t = 200 for the advection equation example presented in Sec. IV A. The graphs indicate that the FC and CL(4) computing-time curves reach the 10−2-accuracy level at about 2.7 s (using 1400 discretization points) and at 1300 s (using 116 000 cells), respectively.

Image of FIG. 4.
FIG. 4.

(Top left) Initial conditions for the velocity u in the one-dimensional (non-linear) Burgers equation example presented in Sec. IV. (Top right) Comparison of the corresponding highly resolved, converged FC solution at time t = 30 (solid line, obtained numerically, as indicated in the text, with a relative accuracy of less than 10−5) and the corresponding FC solution (shown in black dots), obtained using a 11 500-point discretization. The relative maximum error in the latter FC solution is 1.2 × 10−2. (Bottom left and right) Errors vs computing time and discretization size respectively at time t = 30 for the Burgers equation example presented in Sec. IV. The FC and CL(4) computing-time curves reach the 10−2-accuracy level at 106 s (using 12 100 discretization points) and at 2700 s (using 138 000 cells), respectively.

Image of FIG. 5.
FIG. 5.

(Color online) (a) An illustration of the domain decomposition used for a parallel FC solver. The dashed lines represent processor boundaries. In order to update the solution values in the central subdomain (represented by dark circles), the associated processor receives a thin layer of shared values (represented by light circles) from each of four neighboring processors. (b)The treatment of the mirror points across the line of symmetry x = 0. Values are copied from the mirror image values (on the same processor) above the axis according to the symmetry relations (20)–(22). Additionally, the vertical component of the velocity is set to 0 on the axis as required by (22). (c) A portion of a grid covered by overlapping curvilinear patches. (d) As indicated by solid filled regions, one of the patches subdivided into subpatches for distribution in a parallel computing environment.

Image of FIG. 6.
FIG. 6.

A circular arc transducer with focal length F, aperture radius a and radius of curvature .

Image of FIG. 7.
FIG. 7.

(Color online) Stages of the FC focused transducer solver. In Stage I, the computational window is fixed in space next to the transducer (left image). The computation proceeds in Stage I until the signal is centered within the computational window (center image). At this point, the solver enters Stage II computation and the computational window begins moving away from the transducer (right image).

Image of FIG. 8.
FIG. 8.

Comparisons of the KZK and Navier-Stokes (DNS) solutions for the 5λ aperture radius transducer. (a) Peak positive on-axis pressure. (b) Peak negative on-axis pressure. Axial pressure waveform (c) 50λ and (d) 80λ from the transducer.

Image of FIG. 9.
FIG. 9.

(Color online) Comparisons of the amplitudes of the first five harmonics (top to bottom) for the 5λ aperture radius transducer. (Left) Results obtained from the Navier-Stokes FC simulation. (Right) Results obtained from the KZK model. Horizontal axes are measured in focal lengths. Vertical axes are measured in aperture radii.

Image of FIG. 10.
FIG. 10.

Additional comparisons of the KZK and Navier-Stokes (DNS) solutions for transducers with various aperture radii. Peak positive on-axis pressure for the (a) 10λ, (b)20λ, and (c) 30λ aperture radius transducers. (d) Axial pressure waveform 50λ from the transducer for the 30λ transducer.

Image of FIG. 11.
FIG. 11.

(Color online) Comparisons of the amplitudes of the first five harmonics (top to bottom) for the 30λ aperture radius transducer. (Left) Results obtained from theNavier-Stokes FC simulation. (Right) Results obtained from the KZK model. Horizontal axes are measured in focal lengths. Vertical axes are measured in aperture radii.

Image of FIG. 12.
FIG. 12.

(Color online) Pressure field for focusing transducer simulations. (a) Propagation and focusing in a homogeneous medium. (b) Propagation and focusing across an array of scatterers.

Loading

Article metrics loading...

/content/asa/journal/jasa/132/4/10.1121/1.4742722
2012-10-03
2014-04-19
Loading

Full text loading...

This is a required field
Please enter a valid email address
752b84549af89a08dbdd7fdb8b9568b5 journal.articlezxybnytfddd
Scitation: Fourier continuation methods for high-fidelity simulation of nonlinear acoustic beams
http://aip.metastore.ingenta.com/content/asa/journal/jasa/132/4/10.1121/1.4742722
10.1121/1.4742722
SEARCH_EXPAND_ITEM