The use of a parametric array source and nearfield scanning in the characterisation of panel materials for underwater acoustics

Lian Wang, Stephen Robinson, Victor Humphrey

National Physical Laboratory,

Hampton Road, Teddington, UK

ABSTRACT

The properties of the materials used in underwater acoustics are important for applications such as acoustic windows, reflectors and baffles, acoustic barriers or screens, decoupling materials, and anechoic coatings. To characterise the performance of such materials at frequencies above 1 kHz, measurements are typically undertaken on samples of the material in the form of finite sized panels. Such measurements suffer from uncertainty due to the finite size of the panel (leading to contaminating signals from edge diffraction), and the difficulty in simulating the ideal plane-wave insonification. This paper describes work at the UK National Physical Laboratory to minimise these effects by use of: (i) a parametric array as a sound source that provides a directional beam and short broadband pulses; and (ii) nearfield scanning using a hydrophone to sample the complex sound pressure field interacting with the test sample, decomposing the sound field into its plane-wave components. Results are presented of these techniques applied to measurements in laboratory test tanks at frequencies between a few kilohertz and a few hundred kilohertz to determine the reflection and transmission performance of a range of test samples, including panels consisting of homogeneous polymers and materials with regular periodic structure.

1. INTRODUCTION

The acoustic properties of materials that are used for underwater applications are routinely determined by measurements of insertion loss and reflection loss made on test samples consisting of planar (flat) panels of the material. The measurements are commonly made in a test tank, where hydrophones are typically placed in front and behind the panel to determine the reflected and transmitted acoustic fields, with the incident field being measured without the test panel present. Ideally, to accurately determine the response of the material in this form requires that:

i. The test panel is large so that the measurements are not corrupted by diffracted signals from the edges of the sample; ii. The measurements are made in an acoustic free-field with no contamination by reflections from boundaries (such as tanks walls); iii. The test sample is insonified by a plane-wave from a specified direction; iv. The measurements are made at a range such that nearfield effects are not important.

To minimise the effect of the contaminating signals outlined in (i) and (ii), it is beneficial to utilise short acoustic pulses such that the desired signals (incident, reflected and transmitted) may be isolated and measured by time-gating techniques (as the contaminating signals follow

a longer path and arrive later in time). In addition, to minimise the amplitude of the diffracted signals, it is desirable to take advantage of directivity in the source and receiver. A directional source potentially provides reduced insonification of the test panel edges, and a directional receiver can be used to discriminate against signals arriving at high incidence angles. At NPL, a parametric array is used as a sound source to provide a directional beam and short broadband pulses [ Humphrey, et al, 2008] . The parametric array provides some source directivity to mitigate the effect of reflected signals from the test tank, test panel support structure and diffracted signals from the panel edges. However, diffracted signals from the panel edges can still be problematic at low kilohertz frequencies for panels of the order of 1 m by 1 m in size. To improve performance, a directional receiver is also employed in the form of an eight-element pseudo random hydrophone array. With this arrangement, it is possible to conduct measurements down to as low as 1 kHz (depending on panel performance). In Section 2, a description is provided of the acoustic source, the single hydrophone and array receivers, the experimental set up and results for a test panel [Wang, et al, 2014, Beamiss, et al, 2015 ].

Although the above techniques help in the testing of homogeneous panels at normal and oblique incidence, the lack of plane-wave conditions can influence the results due to the incident field containing waves with a range of incident angles [ Humphrey, 1985, Humphrey, 1986, Humphrey, et al, 2008 ]. In addition, if the test panel is not homogeneous and contains internal structure, even the normal incidence results obtained a hydrophone placed close to test panel may give different results depending on the hydrophone position. These latter issues (outlined in (iii) and (iv) above), can be addressed by nearfield scanning using a hydrophone to sample the complex sound pressure field interacting with the test sample, and then decomposing the sound field into its plane-wave components [Humphrey 1986, Williams, 1999]. Results are presented here showing these techniques applied to measurements in laboratory test tanks to determine the reflection and transmission performance of a test sample with regular periodic structure.

2. PARAMETRIC ARRAY SOURCE AND DIRECTIONAL RECEIVER

A parametric array is used as the acoustic source for the work described here [Humphrey and Berktay, 1985, Humphrey, 2009, Westervelt, 1963]. Such a source uses the non-linear propagation of primary wavefields to generate additional lower frequency (secondary) components that are then used to insonify the test panel. Such a source has the advantage of producing a more directional sound field than would be possible from a linear source of similar size. In the arrangement used for these measurements, the primary transducer is driven with a short pulse of 300 kHz carrier frequency, with a raised cosine bell envelope. The low-frequency secondary waveform generated on axis can be shown to be proportional to the second derivative, with respect to time, of the square of the transmitted pulse envelope [Humphrey, 2009]. The generated waveform shape and spectrum is then easily modified by altering the envelope function. In practice, for parametric array measurements in confined spaces, it is necessary to limit the length of the interaction region of the primary beams, which is achieved by placing a panel of absorbing material, known as the acoustic filter, across the field at some distance from the source transducer to absorb the high-frequency primary beams and transmit only the low-frequency beam [Humphrey, 2009].

The test facility used for the measurements consists of a cylindrical tank of external dimensions 7.6 m long and 2.5 m in diameter; the tank is capable of being pressurized to simulate increased

water depth up to a maximum hydrostatic pressure of 68 bar. The facility also allows the water temperature to be controlled in the range from 2 ºC to 35 ºC. There are two access ports, the centres of which are 2.4 m apart. A diagram of the arrangement inside the vessel is shown in Figure 1. The parametric array primaries are generated by a 300 kHz piezoelectric transducer placed at the end of the vessel, with array truncation provided by the acoustic filter (a 35 mm thick Expancel-filled polyurethane sheet). The simplest method of capturing the energy incident upon, reflected from and transmitted by the test panel is via a single hydrophone placed on either side of the test object. Spectral analysis is then performed on the time histories, from which the reflection and transmission loss can be calculated from the incident, reflected and transmitted pressure waves. For reflection measurements at low kilohertz frequencies it is necessary to subtract from the reflection time history the incident signal recorded at the same point and under the same conditions without the test panel present (because the incident and reflected signals overlap in time).

Figure 1: Schematic diagram of the measurement configuration in the NPL Acoustic Pressure Vessel (APV), showing the location of transducer, acoustic filter, test panel and receive array.

The acoustic pressure wave generated by the truncated parametric array can be readily acquired and time resolved by a single hydrophone under free-field conditions. However, the benefit obtained by additional directivity from the parametric array source is substantially reduced at frequencies below 10 kHz. Since the signals of interest are plane waves travelling in directions orthogonal to the panel surface, and the contaminating signals are predominately travelling in directions across the panel at significant angles (from panel edges or vessel walls), it is possible to utilise a directional planar array to select the plane wave components of interest [Beamiss, 2015, Wang, et al 2017]. The planar array used was constructed from eight Neptune Sonar T293 spherical hydrophones of 20 mm diameter, which were mounted using a mount which was acoustically transparent at frequencies below 10 kHz (consisting of thin 0.6 mm diameter cord and a 550 mm square aluminium frame). The configuration was a pseudo random, 2D arrangement covering an area of approximately 0.4 m 2 . Measurements were performed by amplifying the hydrophone outputs using an ETEC B2008 eight channel charge amplifier and simultaneously capturing the signals using a National Instruments multifunction DAQ using

bespoke LabView software. In the analysis, an average was initially calculated across the eight channels and the resulting waveforms averaged over a number of pulses. Measurements were conducted on a polymer test panel designed to have suitable performance in the frequency range 1 kHz to 15 kHz. This was 800 mm wide, 900 mm high and 125 mm thick and consists of a two-layer acoustic absorber fabricated from polyurethane. The results for reflection loss and transmission loss are shown in Figures 2 and 3 below. The thickness mode resonance of the test panel is clearly visible in the results, and the change in the behaviour as a function of hydrostatic pressure (immersion depth) and temperature is evident, with results obtained down to 1 kHz.

Refecion Loss) Frequency (kH2) mC eae

Figure 2. Results for the reflection loss of the test panel obtained using the receive array for water temperatures of 8 °C and 20 °C at atmospheric pressure (left), and for hydrostatic pressures of 0.1 MPa, 0.7 MPa, and 1.4 MPa at a temperature of 20 °C (right).

LNPL2: 20°C: Ske puto

Figure 3. Results for the transmission loss of the test panel obtained using the receive array for water temperatures of 8 °C and 20 °C at atmospheric pressure (left), and for hydrostatic pressures of 0.1 MPa, 0.7 MPa, and 1.4 MPa at a temperature of 20 °C (right).

ranemssion Loss (28) ry

3. NEARFIELD SCANNING

Figure 4 shows the diagram of the measurement setup used for the nearfield scanning work. In this case a higher frequency source transducer, with a diameter of 44 mm and a resonance frequency 1 MHz (made by Precision Acoustics), was used to generate the parametric array. The transducer was driven with a 1 MHz sine wave amplitude modulated with a raised cosine

‘Transmission Loss (48) INPLE: 20°C: 8 ke pao

bell envelope based on a 60 kHz signal; this generated a Ricker pulse secondary signal in the interaction region. To avoid any secondary signal generation by the test panel and hydrophone the primary field was terminated by an acoustic filter.

A Reson TC4035 hydrophone was used to measure the resulting broadband pulse close to the test object. This hydrophone was chosen for its small size (a cylindrical element of 1.5 mm diameter and 1.5 mm length); this enabled the desired scan resolution to be achieved without significant spatial averaging by the hydrophone element. The signal from the hydrophone was then captured by a National Instruments PXI-5922 digitiser and saved as an ASCII data file. The NPL small open-tank facility (SOTF) was used for the scan test. It consists of a 3.5 m x 1.5 m x 1.5 m GRP tank filled with fresh water. It has two precision positioning carriages, both of which have movement in four axes (X, Y, Z and azimuth rotation, with 10 µm linear resolution and 0.01° in rotation). These are positioned under full computer control, with software which enables linear and planar scans for acquisition of complex acoustic pressure (magnitude and phase). The test panel used had a periodic structure and consisted of stainless steel tubes which were air-filled and were 620 mm high, 6 mm in diameter, with 0.5 mm wall thickness. There were 20 tubes with a separation of 20 mm in the test object. This will be referred to as the ‘rods’ test panel.

Figure 4: Schematic diagram of measurement configurations for planar scans in planes parallel to the test object (left) and the test panel made of a grid of parallel stainless-steel tubes (right).

Line scans were undertaken to measure the forward and backward scattering from the test panel for normal incidence or for off-normal incidence with an incident angle of 15 o to the panel normal. Two types of line scan were used as shown in Figure 5, one with the scan line normal to the incident wave axis and the other with the scan line parallel to the panel. The length of the scan lines for the tests presented in this paper was 600 mm with a step increment of 1 mm.

The advantage of the scan line normal to the incident wave axis is simplicity with only motion along the y -axis required for the carriage supporting the hydrophone. However, there is a limit

on the maximum incidence angle at 45 o that can be used for back scattering measurements with this approach.

The acoustic response of the rods panel was characterised using the near-filed scan approach in the SOTF. The signal pulse with the panel present was measured at pre-determined positions along a scan line and stored in a 2D data array of time and position. This process was then repeated without the panel to obtain the incident field. The signal with panel contains response of the panel to the incident acoustic wave, while the signal with only incident wave provides a reference that the panel response can be assessed against. The frequency range used was from 20 kHz to 300 kHz.

Figure 5: Schematic of measurement technique showing two scan line directions for near- field measurements: normal to incident wave axis (red dots) and parallel to test object (green dots).

The measured signal pulses for this experiment were very short due to the higher frequency band, so it was possible to use time-gating to remove multipath interferences due to any reflections from the water surface, tank walls and bottom. An FFT was then applied in the time and then spatial domains to convert the signals into results as a function of frequency and angle. Some measured results are presented here in both the conventional Cartesian coordinate system and also a polar coordinate system, with the radial coordinate representing frequency.

Figure 6 shows the results for the back-scattering signal from the rods panel measured by the near field scan as a function of angle and frequency with a normal incident wave to the panel (top plot), and off-normal incident wave at 15 o for the middle and bottom plots. The middle plot is for the scan line normal to incident wave, and the bottom is for scan line in parallel with the panel. The results are normalised with the plane-wave component derived from the reference signal. The results are shown in Cartesian coordinates on the left, and polar coordinates on the right of the figure. The back-scattering from the rods consists of a main

beam at the centre of angle, and three curved grating lobes on each side of the main beam in Cartesian coordinates. The positions of the grating lobes are determined by frequency, distance between two adjacent rods and angle. The grating lobes become parallel lines with an equal distance between the lines in the polar coordinates, and rotate with the same angle shift of the main beam if the incident wave is oblique. It can be seen that there is good agreement between the measured and predicted grating lobes in black dash line. It can be seen that the signal is quite noisy from frequency above 250 kHz especially for the bottom one as signal to noise ratio decreases with frequency at the high end of the band. It may also be observed that the angle of the main beam is different for the two scan lines. It is 30 o for scan line normal to incident wave, and 15 o for scan line in parallel with the panel as shown in Figure 5. The angle range covered by scan line in parallel with panel is greater than that by the other scan line. There is clearly spreading in angle of the main beam and grating lobes due to the curvature of the incident wave from a source at a finite distance.

Figure 8 shows forward-scattering signal measured in near field from rods panel as a function of angle and frequency with a normal incident wave to the panel (top), and off-normal incident wave at 15 o for the middle and bottom plots. The structure of the main beam and grating lobes for the rods panel in the top plot is the same as that for the back-scattering case in Figure 7. The level of the main beam is much higher for the forward-scattering due to the spare distribution of the rods. The level of the grating lobes is comparable between the back- scattering and back-scattering for rods panel. The main beam stays at the centre for forward scattering with the panel and the scan line normal to the acoustic axis of the incident wave.

4. SUMMARY

Two applications to characterise acoustic properties of materials in the form of finite sized planar panels under laboratory conditions have been demonstrated using a parametric array source and directional receive arrays. The first one uses a directional receiver consisting of a random planar array with 8 hydrophones which has been used for testing a homogenous polymer panel where the reflection and transmission in the normal direction of the panel were measured at frequency of low kilohertz frequencies. In the second, near-field scans were carried out to form a synthesised line array across a test panel to assess the spatial response of the panel which exhibits periodic structure and generates a grating lobe response in different wave directions. The results demonstrate the efficacy of the methods to characterise test panels and overcome the limitations inherent in such measurements due to the finite size of the test panel and the lack of idealised acoustic fields (for example, non plane-wave insonification). The methods offer the ability to characterise the performance of panels over a wide range of acoustic frequencies and to determine the spatial response as a function of angle.

ER vs frequency and angle (fft): BS rods Odeg oy (kHz) ‘Amplitude (48) ‘Amplitude (dB) 8 Angle (degs)

ER vs frequency and angle (fft): BS rods 15deg ‘Amplitude (48) 8 Angle (degs)

Figure 6: Normalised back-scattering signal measured in near field as a function of angle and frequency with rods panel normal to incident wave (top), 15 o off-normal (middle and bottom). The scan line is normal to incident wave for the middle plot and parallel with panel for the bottom plot.

‘Amplitude (48) ‘Amplitude (dB) Angle (degs)

IL vs frequency and angle (fft): FS rods Odeg ° 20° ‘Amplitude (48) ‘Amplitude (dB) 8 Angle (degs)

IL vs frequency and angle (fft): FS rods 15deg ‘Amplitude (48) ‘Amplitude (dB) 8 Angle (degs)

Figure 7: Normalised forward-scattering signal measured in near field as a function of angle and frequency with rods panel normal to incident wave (top), 15 o off-normal (middle and bottom). The scan line is normal to incident wave for the middle plot and parallel with panel for the bottom plot.

ILvs frequency and angle (ft): rods 1Sdeg IL off normal ILvs frequency and angle (fft): rods 15deg IL off normal 2 cA ‘Amplitude (48) ‘Amplitude (dB) 8 -50 0 50 Angle (degs)

1. REFERENCES Humphrey V. F., Robinson S. P., Smith J. D., Martin M. J., Beamiss G. A., Hayman G. and Carroll N. L. “Acoustic characterisation of panel materials under simulated ocean conditions using a parametric array source”, J. Acoust. Soc .Am. vol 124, p.803-814, 2008. Humphrey V. F., “The measurement of acoustic properties of limited size panels by use of a parametric source”, J. Sound. Vib., vol. 98, p67-81, 1985. Humphrey, V. F. “The influence of the plane wave spectrum of a source on measurements of the transmission coefficient of a panel,” J. Sound Vib., vol. 108, 261–271, 1986 Humphrey, V.F. and Smith, J.D. “Panel transmission measurements: the influence of the non- plane wave nature of the incident field”, Proceedings of Acoustics’08, p.3671-3676, Paris, 2008. Humphrey, V. F., and Berktay, H. O. “The transmission coefficient of a panel measured with a parametric source,” J. Sound Vib. 101, 85–106, 1985. Humphrey V. F., S. P. Robinson, G. A. Beamiss, G. Hayman J. D. Smith, M. J. Martin, and N. L. Carroll. “Sonar material acoustic property measurements using a parametric array”. J. Acoust. Soc. Am. vol 125, p2718, 2009. Beamiss, G.A., Robinson, S.P., Wang, L.S., Hayman, G., Humphrey V. F., Smith J. D., Martin, M.J. “The use of a sparse planar array sensor for measurement of the acoustic properties of panel materials at simulated ocean conditions”, Proceedings of UACE2015 - 3rd Underwater Acoustics Conference and Exhibition, Crete, p 235-242, June 2015. Wang, L., Robinson, S.P. and Hayman, G. “Assessment of measurement performance using a random planar array on material panels”, Proceedings of 1st International Conference and Exhibition on Underwater Acoustics, Greece, ISBN: 978-618-80725-0-3, p 1169-1175, 2014. Wang, L., Robinson, S.P. and Hayman, G. “Assessment of measurement performance using a random planar array on material panels”, Proceedings of 1st International Conference and Exhibition on Underwater Acoustics, Greece, p 855-863, 2017. Westervelt, P. J. “Parametric acoustic array,” J. Acoust. Soc. Am. vol. 35, p. 535–537, 1963. Williams, E.G. “Fourier Acoustics – Sound Radiation and nearfield holography”, Academic Press, 1999.