Model-Based Design of Low Frequency Lamb Wave EMATs for Mode Selectivity

Abstract

A low-frequency, omni-directional A0 Lamb wave ElectroMagnetic Acoustic Transducer (EMAT) is developed for applications in guided wave tomography, operating at 50 kHz on a 10 mm thick steel plate. The key objective is to excite an acceptably pure A0 wave mode in relation to the S0 mode, which can also be present at this operating point and is desired to be suppressed by approximately 30 dB. For that, a parametric Finite Element (FE) model of the design concept is implemented in a commercially available FE software, where the bias magnetic field is calculated initially, then combined with the eddy current caused by the induction coil to produce a force. A numerical optimization process employing a genetic algorithm is set up and the EMAT design is optimized to yield an improved A0 mode selectivity. The parameters subjected to optimization are the magnet diameter and the magnet lift-off, which control the direction of the exciting force in the skin depth layer and therefore the mode selectivity. Although there are three possible electromagnetic acoustic interaction mechanisms, the optimisation considers only the Lorentz force, as its performance surface contains a clear optimum and from the optimised design a physical prototype is built. The FE model is validated against measurements on an aluminium plate for the Lorentz force excitation mechanism and on a steel plate for both the Lorentz and magnetisation force. For the steel plate, it is found that only considering the Lorentz force leads to a significant overestimation of the mode selectivity, as the S0 amplitude is underestimated by the Lorentz force, but the A0 amplitude remains mainly uninfluenced. Further, it has been found that additionally including the magnetisation force into the optimisation leads to a better mode selectivity, however, the optimisation drives the optimum to a minimum magnet diameter and therefore reduces the EMAT sensitivity. In a numerical study robustness is shown for fairly large variations of the magnet lift-off and the magnetic permeability. Based on the findings, a two-step model-based design approach is proposed whereby only the Lorentz force is used in the first step for the optimisation and then in a second step, a realistic estimate of the mode selectivity of the optimised design can be obtained by additionally considering the magnetisation force.

Accelerated optimizations of an electromagnetic acoustic transducer with artificial neural networks as metamodels

“…One design variable and one objective function were used in this optimization, accomplished only through observation of a set of curves corresponding to different design variables, instead of using a real optimization algorithm. A spiral coil EMAT was optimized using genetic algorithm optimization procedure in the global optimization toolbox of Matlab in Seher et al (2014Seher et al ( , 2015. The ratio of the amplitude of the A0-mode Lamb waves to that of the S0 mode was selected as the objective function to be maximized, i.e., preferably generating the A0 mode.…”

“…For the EMAT on the steel plate, n is chosen to be 1, i.e., R C = λ 4 , similar to Seher et al (2015).…”

On the separation of Lorentz and magnetization forces in the transduction mechanism of Electromagnetic Acoustic Transducers (EMATs)

“…In the following numerical simulations what could be called a strongly paramagnetic material model will be used that assumes that the magnetization  MH is linearly proportional to the magnetic field H, where χ denotes the magnetic susceptibility. This simplistic linear magnetization model is identical to the one recently used by Rouge et al [23,24] and Seher et al [29] and is well suited to create benchmark problems for the comparison of different EMAT models. In the following the relative permeability r 1     will be used as the principal material parameter to characterize the ability of the material to magnetically polarize and the magnetic flux density will be written as…”

“…Jafari-Shapoorabadi et al [22] and Rouge et al [23][24][25] followed essentially the same approach of calculating the total Lorentz body force according to f L3 while combining it with the magnetization force f m2 and f m3 , respectively, but without any surface traction component. Ludwig [26,27]  TT to study the mode selectivity of Lamb-type guided waves in ferromagnetic plates [29]. Whether or not these models provide good approximations of the EMAT's performance depends on the type of EMAT configuration and the properties of the ferromagnetic material the EMAT is applied on.…”

Computational Time Reversal for NDT Applications Using Experimental Data

“…A different class of solution methods is based on probabilistic analysis, as in [12]. Identification based on the spectrum (modal analysis) also constitutes an important class of methods; see, e.g., [13][14][15][16]. An alternative to these methods is the use of Arrival Time Imaging, also called Travel Time Migration or Kirchhoff Migration [17][18][19].…”