Ultrasonic flow meters (UFMs) based on transducer arrays offer several advantages. With electronic beam steering, it is possible to tune the steering angle of the beam for optimal signal-tonoise ratio (SNR) upon reception. Moreover, multiple beams can be generated to propagate through different travel paths, covering a wider section of the flow profile. Furthermore, in a clamp-on configuration, UFMs based on transducer arrays can perform self-calibration. In this manner, userinput is minimized and measurement repeatability is increased. In practice, transducer array elements may break down. This could happen due to aging, exposure to rough environments, and/or rough mechanical contact. As a consequence of inactive array elements, the measured transit time difference contains two offsets. One offset originates from non-uniform spatial sampling of the generated wavefield. Another offset originates from the ill-defined beam propagating through a travel path different from the intended one. In this paper, an algorithm is proposed that corrects for both of these offsets. The algorithm also performs a filtering operation in the frequency-wavenumber domain of all spurious (i.e., flow-insensitive) wave modes. The advantage of implementing the proposed algorithm is demonstrated on simulations and measurements, showing improved accuracy and precision of the transit time differences compared to the values obtained when the algorithm is not applied. The proposed algorithm can be implemented in both in-line and clamp-on configuration of UFMs based on transducer arrays.

Coatings with dynamic surface structures are appealing to many applications like haptics and soft robotics. Restrictively, the speed of the surface dynamics in these coatings is often limited to frequencies below 1 kHz, which makes them unsuitable for applications like acoustics and communication optics. This work describes a method to create high-frequency surface dynamics controlled by alternating electric fields on a substrate-contact-modulated coating that consists of an elastic poly(dimethyl siloxane) network supported by SU-8 microstructures. The principle is based on the global application of Maxwell stress that is locally resisted by the supporting SU-8 microstructures. In-between the microstructures the elastic material is stretched, causing a large deformation of the surface topography, which is supported by the authors’ finite element method models. By applying a high-frequency alternating field, they discovered resonance effects at frequencies up to 230 kHz, where the surface of the coating vibrates at high speeds and large amplitudes. At these high frequencies, the coatings can produce and detect ultrasound waves underwater, indicating their potential for ultrasound transducers in the future.

In recent years, several fitting techniques have been presented to reconstruct the parameters of a plate from its Lamb wave dispersion curves. Published studies show that these techniques can yield high accuracy results and have the potential of reconstructing several parameters at once. The precision with which parameters can be reconstructed by inverting Lamb wave dispersion curves, however, remains an open question of fundamental importance to many applications. In this work, we introduce a method of analyzing dispersion curves that yields quantitative information on the precision with which the parameters can be extracted. In our method, rather than employing error minimization algorithms, we compare a target dispersion curve to a database of theoretical ones that covers a given parameter space. By calculating a measure of dissimilarity (error) for every point in the parameter space, we reconstruct the distribution of the error in that space, beside the location of its minimum. We then introduce dimensionless quantities that describe the distribution of this error, thus yielding information about the spread of similar curves in the parameter space. We demonstrate our approach by considering both idealized and realistic scenarios, analyzing the dispersion curves obtained numerically for a plate and experimentally for a pipe. Our results show that the precision with which each parameter is reconstructed depends on the mode used, as well as the frequency range in which it is considered.

Common clamp-on ultrasonic flow meters consist of two single-element transducers placed on the pipe wall. Flow speed is measured noninvasively, i.e., without interrupting the flow and without perforating the pipe wall, which also minimizes safety risks and avoids pressure drops inside the pipe. However, before metering, the transducers have to be carefully positioned along the pipe axis to correctly align the acoustic beams and obtain a well-calibrated flowmeter. This process is done manually, is dependent on the properties of the pipe and the liquid, does not account for pipe imperfections, and becomes troublesome on pipelines with an intricate shape. Matrix transducer arrays are suitable to dynamically steer acoustic beams and realize self-alignment upon reception, without user input. In this work, the design of a broadband 37×17 matrix array (center frequency of 1 MHz) to perform clamp-on ultrasonic flow measurements over a wide range of liquids (c=1000-2000m/s, α≤1 dB/MHz · cm) and pipe sizes is presented. Three critical aspects were assessed: efficiency, electronic beam steering, and wave mode conversion in the pipe wall. A prototype of a proof-of-concept flowmeter consisting of two 36-element linear arrays (center frequency of 1.1 MHz) was fabricated and placed on a 1-mm-thick, 40-mm inner diameter stainless steel pipe in a custom-made flow loop filled with water. At resonance, simulated and measured efficiencies in water of the linear arrays compared well: 0.88 and 0.81 kPa/V, respectively. Mean flow measurements were achieved by electronic beam steering of the acoustic beams and using both compressional and shear waves generated in the pipe wall. Correlation coefficients of R2>0.99 between measured and reference flow speeds were obtained, thus showing the operational concept of an array-based clamp-on ultrasonic flowmeter.

Current ultrasonic clamp-on flow meters consist of a pair of single-element transducers that are carefully positioned before use. This positioning process consists of manually finding the distance between the transducer elements, along the pipe axis, for which maximum signal-to-noise ratio (SNR) is achieved. This distance depends on the sound speed, thickness, and diameter of the pipe and on the sound speed of the liquid. However, these parameters are either known with low accuracy or completely unknown during positioning, making it a manual and troublesome process. Furthermore, even when sensor positioning is done properly, uncertainty about the mentioned parameters, and therefore on the path of the acoustic beams, limits the final accuracy of flow measurements. In this research, we address these issues using an ultrasonic clamp-on flow meter consisting of two matrix arrays, which enables the measurement of pipe and liquid parameters by the flow meter itself. Automatic parameter extraction, combined with the beam-steering capabilities of transducer arrays, yields a sensor capable of compensating for pipe imperfections. Three parameter extraction procedures are presented. In contrast to similar literature, the procedures proposed here do not require that the medium be submerged nor do they require a priori information about it. First, axial Lamb waves are excited along the pipe wall and recorded with one of the arrays. A dispersion curve-fitting algorithm is used to extract bulk sound speeds and wall thickness of the pipe from the measured dispersion curves. Second, circumferential Lamb waves are excited, measured, and corrected for dispersion to extract the pipe diameter. Third, pulse-echo measurements provide the sound speed of the liquid. The effectiveness of the first two procedures has been evaluated using simulated and measured data of stainless steel and aluminum pipes, and the feasibility of the third procedure has been evaluated using simulated data.

Acoustic wave propagation in ultrasonic flow measurements is typically assumed to be linear and reciprocal. However, if the transmitting transducer generates a sufficiently high pressure, nonlinear wave propagation effects become significant. In flow measurements, this would translate into more information to estimate the flow and therefore a higher precision relative to the linear case. In this work, we investigate how the generated harmonics can be used to measure flow. Measurements in a custom-made flow loop and simulations using the Khokhlov–Zabolotskaya–Kuznetsov (KZK) equation will show that the second harmonic component provides similar transit time differences to those obtained from the fundamental component, their linear combination results in more precise flow measurements compared to the estimations with the fundamental component alone.

Current ultrasonic clamp-on flow meters are based on single-element transducers that require manual calibration by aligning these to a fixed acoustic path. Moreover, the size and operational frequency of the transducers cannot be adapted to the parameters of the pipe and the liquid, which are in practice not precisely known a priory. A set of two transducer arrays could be used to solve these issues. With an array, properties of the pipe and the liquid can be estimated before measuring flow. Furthermore, electronic beam steering can be used for auto-alignment of the acoustic beam, reducing the need for manual calibration. Moreover, an array allows for the use of signal processing to suppress the effects of spurious Lamb waves propagating in the pipe wall. This research work describes the acoustic design process of a transducer array for ultrasonic clamp-on flow measurements for a wide range of conditions. First, performance requirements are defined. Then, the design models are presented, and a step by step process of the acoustic stack design of the transducer array is described. At each design step, material dimensions are optimized to achieve a thickness resonance mode at 1 MHz within a bandwidth of interest between 0.2 MHz and 2 MHz. Finally, the expected performance of the designed array is reported, based on simulation results.