Page 32 - \Winter 2015

Computer Simulation for
Predicting Acoustic Scattering from Objects at the Bottom of the Ocean
The TS is also strongest (red color) in these two directions because there are parts of the surfaces of the shell that are oriented perpendicular to the incident wave and therefore reflect part of the energy directly backward; at other angles the surfaces are oblique, reflecting energy in other direc- tions. Other intense areas of the template (yellow, green, light blue) are caused by resonances of various types of in- ternal elastic waves.
Acoustic Signature vs. Imaging
The dimensionless frequency band of acoustic signatures is the same for all objects, no matter how large or small: from about ka = 1 to about ka = 30, where ka is dimensionless frequency, k is wavenumber (2л/λ), λ is the wavelength of the insonifying plane wave (recall Figure 2) and a is an ap- proximate “radius” of the object, namely, an average “ra- dius” of chunky objects or shorter “radius” of long objects. At the lower frequencies, wavelengths are comparable to the exterior dimensions of the object; at the higher frequencies wavelengths are comparable to smaller interior dimensions. Equivalently, this is the range of the lower natural modes of vibration of the object. To illustrate, the vertical frequency axis in Figure 3, using the speed of sound in water as 1500 m/s and a = 0.25m, ranges from ka = 1 to ka = 26.
Sonar classification has traditionally relied on much higher frequencies, where wavelengths are very small relative to the object and therefore can produce a rough image of the object, that is, its external shape, but it cannot reveal inter- nal composition because the shorter wavelengths can’t pen- etrate very deeply before they die out due to attenuation per wavelength. Consequently, acoustic signatures and imaging complement each other in modern sonar systems. A key dif- ference between the two approaches is the amount of infor- mation needed to interpret sonar output: imaging produces a picture of the object so an engineer can identify the object simply by looking at the image, whereas a signature produc- es only a TS template, which requires a computer to detect meaningful patterns in the template.
The Need for Computer Simulation
One approach to the computer search for such patterns is to see if there are similarities with the templates of similar ob- jects in a variety of realistic configurations, for example, rest ing on the bottom, partially buried, fully buried, tipped, etc., or in different types of sediment such as sand, clay, mud, etc., as well as objects with similar construction, for example, de- coys or manufacturing variations. All of these variations can
have a significant effect on the vibrational response of the object and hence its acoustic signature. That requires having a large library of reference acoustic signature templates.
To construct such a library, one could perform experiments on actual objects. But experiments are expensive and time consuming so only a few can be performed, and one can- not perform experiments on unavailable objects or environ- ments. Computers, however, can model virtually any object/ environment scenario of interest, including nonexistent sce- narios. The cost of computer resources per model is negli- gible compared to that of a real underwater experiment and often faster by orders of magnitude, sometimes enabling hundreds or thousands of templates to be computed in the same time as performing one underwater experiment. There is clearly a need for a computer simulation system that is both high-fidelity and computationally fast.
Computer Simulation
The Naval Surface Warfare Center Panama City Division (NSWC PCD) has developed a high-fidelity, broadband (CW analyses), three-dimensional (3-D), finite-element (FE) computer simulation system called PC-ACOLOR (Panama City-Acoustic COLOR), which models the acous- tic scattering signature (also called acoustic color) of single or multiple realistic targets at the bottom of the ocean.
The principle challenges to developing such a system were as follows:
• Multiscale spatially: From small details in the objects (cm)
to large distances in the ocean (km).
• Broadband: A five-octave range, ka ≈ 1 to 30.
• A need for extraordinarily high computational efficiency:
One acoustic signature template requires sweeping typi- cally over several hundred frequencies, and, for each fre- quency, several hundred aspect angles, requiring O(105) 3-D models. Modeling techniques developed by the author since the 1980s have enabled the code to currently compute about one template per day, an adequate pace for applica- tions so far. Nevertheless, the next level of applications will require computing several hundred templates per day in order to create statistically robust acoustic signature librar- ies. R&D to achieve such speeds is underway and should be operational when this article is published (see The Way Forward at the end of this article).
The next three subsections describe the approach used by NSWC PCD: the physics, the governing mathematics, and the computational modeling technique.
30 | Acoustics Today | Winter 2015