 Home |
|  Home |
1118KB
|
ULTRASONIC PHASED ARRAY AND SYNTHETIC APERTURE IMAGING IN CONCRETE K.J. Langenberg1, K. Mayer1, R. Marklein1, P. Ampha1, M. Krause2, F. Mielentz2 1 Department of Electrical Engineering and Computer Science, University of Kassel, 34109 Kassel, Germany; 2 Federal Institute for Materials Research and Testing, Unter den Eichen 87, 12205 Berlin, Germany Abstract: For ultrasound in the hundred kilohertz regime concrete is a very inhomogeneous propagation material. The prescribed definition of aggregates with regard to the respective percentage, the size distribution and the material composition together with the percentage of air inclusions can be realized as a computer model to serve as a platform for elastic wave propagation simulations. As a simulation tool we have developed the Elastodynamic Finite Integration Technique (EFIT) in 2D and 3D, thus being able to visualize the radiation field of either conventional transducers, ultrasonic phased arrays and arrays of point contact transducers. Through the measurement of the displacement on the opposite side of a specimen's excitation surface with a laser vibrometer the simulated data can be related to experiments thus clearly predicting the performance of these devices in the concrete environment. A particular NDT problem in concrete consists in the identification of grouting defects in tendon ducts via the application of imaging algorithms like SAFT (Synthetic Aperture Focusing Technique) and its diffraction tomographic Fourier transform version FT-SAFT. EFIT is able to produce synthetic data for a given situation thus providing a test bed for SAFT in concrete to reveal its potential for these applications. Various examples will be given and compared with pertinent experiments. Furthermore, EFIT combined with SAFT serves as a tool to predict the behavior of specimens to be fabricated before they are actually realized. Introduction: For all applications of remote sensing the phased array concept is very prominent because of its fast or real-time applicability and the high capabilities in respect to lateral and axial resolution. Up to now the application of this method for ultrasonic testing in concrete is not widely used because of the extraordinary expense and the rough environment in which the equipment has to be used. With the knowledge about the problems in non- destructive testing of concrete by ultrasonic waves it may be worth to think about it. The wave propagation in concrete material is dominated by a high degree of multiple scattering and high attenuation because of the heterogeneous structure of the "background" material which consists of three components: aggregate, cement, and pores distributed in a more or less statistical manner, roughly characterized by a grading curve for the size of the aggregate which itself consists of sand and gravel stones. The size of the material inhomogeneities recommends the use of ultrasonic waves in the 50 kHz to 200 kHz regime. Together with the problem of coupling this yields a size of our array with 10 elements in a line consisting of 34 mm commercial broadband transducers of about 34 cm. Phased arrays consist of a collection of similar transducers which are fed by a set of pulsers which can fire an impulse or a predefined signal with an appropriate delay and amplitude to reach a desired beam direction and focusing. Different constellations have been discussed in [1]. Because of the really strong behavior of waves in concrete the outcome of investigations with sophisticated techniques is more or less unpredictable. Therefore, empirical investigations in connection with modeling techniques have to be performed. Considering the use of measured data for further processing like Diffraction Tomography or the Synthetic Aperture Focusing Technique (SAFT) the whole cycle of analysis has to be repeated for any new detection problem. In this paper the methods are presented to apply phased array techniques for detecting voids in tendon ducts, which are partially filled with steel cables and mortar. Results: 1. Modeling The calculation of the wave propagation in the given complex structure is only possible by numerical methods. Based on the governing equations of linear elastodynamics a numerical modeling tool was developed and denoted as Elastodynamic Finite Integration Technique (EFIT) [2,3]. It has been validated against many analytical and experimental methods. Even the application of EFIT to strongly inhomogeneous structures like concrete has been investigated successfully [4, 5]. As a model for the medium serves an accumulation of ellipses with different orientation, size and material property, which is generated by a computer code according to statistical specifications and grading curves. This is possible for two and three dimensions. The pores developing during drying of the concrete in the range of 0 to about 0.1 mm are taken into account by grid cells with material properties of air in the interspace of the ellipses which is filled with mortar otherwise. Figure 1 shows a picture of a concrete specimen and the 2D and 3D presentation by the grid generator. Figure 2 shows the wave field (a snapshot at a certain time) which is transmitted by a transducer at the top of a 2D test geometry. Beside the propagation in a homogeneous isotropic background material (mortar) the wave in a concrete composition is shown. The numerical effort of EFIT is predetermined by the discretisation of the area or the volume, the number of time steps which is given by the way the wave has to cover. Additionally, the kind of the measurement which should be modeled by the simulation determines the effort. For monostatic measurements - this means that the position of the transmitting and receiving transducer is common and is changed for every pulse - a complete simulation is necessary for every transducer position (A-scan). The numerical effort is distributed by an extra MPI implementation for distributed computing on a compute cluster of 16 single nodes in our department. Figure 3 shows the used segmentation model and an example of the propagation of a plane wave in a cube of concrete with the edge length of 51 mm. 1
2
3
4
5
6
Process / CPU
8
Domain Decomposition
1 2
4
5
3
6
7
8
7 For experimental studies the Federal Institute for Materials Research and Testing (BAM) provides a phased array arrangement according to Figure 4. Two realizations have been worked out: transducers in a line and transducers in closed package along a line. The first one could be modeled approximately by a 2D EFIT simulation. To reach that, the grid nodes at the top of the test region have been applied by an excitation pulse, whose amplitude and time delay are given in Figure 5. It can be recognized, that the amplitude weighting is given as a Tukey window and the time delay is implemented according to the desired beam steering and focusing but is constant within an array element. Figure 6 gives the results of simulation for unfocused excitation. At first in homogeneous material: for an excitation impulse (100 kHz RC2 (raised cosine, 2 cycles)) a 0°, respectively an 11° experiment is simulated. The displayed point in time shows the plane pressure (P)-wave front. The shear (S)-wave front which propagates with about half the velocity has a very low amplitude, because the S-wave field of a single transducer exhibits a zero line along the acoustical axis (thus into the direction of the ultrasonic beam in the upper row of the figure). Following the maximum amplitude of the wave field during the propagation, one recognizes the beam pattern of the ultrasonic beam (right). This is explicitly seen in Figure 7, which gives a snapshot and a beam of a focused and steered wave field. The properties of the wave field for a prescribed concrete simulation with a grading curve A16 (16 mm maximum aggregate size) and 1% air in a 2D experiment is given exemplarily in Figure 8. From that it is evident, that the beam is attenuated considerably, the focusing is conserved, and behind the wave front a wave field concentration appears which propagates slower than the hypothetic shear-wave front. . Figure 4: Phased array arrangement at the Federal Institute for Materials Research and Testing (Berlin). This effect can be observed better in animations of the wave propagation and is increasing with a higher content of air in the model. To compare the experiment and the simulation, the particle displacement at the bottom (backside) of the specimen was measured (respectively stored during simulation). The complete received signal of the simulation of a focused beam with 11° steering is displayed in Figure 9 (left). At the right side the characteristic of the maximum amplitude along the scanning line is given for the experiment and the simulation. 2. Modeling and Imaging We have seen, that the wave propagation in concrete is not predictable in detail because of the enormous number of scattering events. Only the principle wave front keeps stable to some degree. Therefore it is questionable how apparently sensitive signal processing methods like modern imaging algorithms react on that kind of signals. The combination of modeling tools like EFIT with imaging algorithms like SAFT (Synthetic Aperture Focusing Technique) or FT-SAFT (Fourier Version of SAFT for plane surfaces) [6] based on Diffraction Tomography is of particular importance, because one can examine all effects appearing with imaging algorithms with relevant examples without the need of building expensive concrete specimens. Also the selection of appropriate specimens can be facilitated. A very important task in civil engineering is the non-destructive testing of the interior of tendon ducts, because this characterizes the state of preservation of many expensive buildings like bridges. Tendon ducts contain steel cables, which can be tensioned after hardening of the concrete, as pretensioning of the concrete because this material is not very stable against traction in any form. To prevent the steel cable from corrosion the tendon ducts are then grouted with mortar. Therefore, the state of grouting is a very important quantity in non-destructive testing of concrete. By the way, the position of the tendon duct usually is not well known and has do be determined by means of Ground Penetrating Radar (GPR) methods, where the same imaging algorithms are applied mostly preparing a measurement. a)
transducer array
laser scanning or EFIT-sampling
c)
plitude in a. u.
A
m
1
0,8
0,6
0,4
0,2
0
-200 -100 0 100 200
s in mm But as these tendon ducts consist of steel, electromagnetic waves cannot intrude into them, in contrast to elastodynamic waves. Figure 10 shows results of investigations which should discover the conditions of detectability of grouting errors in tendon ducts. For that, a model akin to Figure 1 was selected, additionally the geometry of a duct with tension cables and mortar with varying grouting was blended in. As original for that arrangement served a photography of a partially filled tendon duct of a specimen at the BAM. Figure 10 (top left) shows the photography, the center displays one realization of the simulated geometry, and in the left figure the snapshot of a wave propagation emitted by a transducer at top edge of the simulated region is shown. To get the synthetic data of a monostatic experiment, the particle velocities under the transducer/receiver probe are integrated with a weighting function and this procedure has to be repeated for 160 transmitter/receiver positions. The so captured data are comparable with experimentally achieved B-scans and can be processed with SAFT under the assumption of a constant average velocity of the medium. The series in Figure 11 shows the reconstruction of different configurations which make the detection of simple voids of air in the two-dimensional case at least doubtful, considering the case of further unknowns. Reconstruction of 2D scanning apertures at comparable specimens show more promising results. To substantiate this, in the following example the incident wave field of one transducer is replaced by the already presented ultrasonic phased array to impinge a plane wave on the region of interest. As figure 12 shows, the information of the already known scattering centers as imaged by FT-SAFT (Figure 11) gets another weighting. But no real knew knowledge about the interior structure of the duct is gained. Acknowledgement: The support of this study by the Deutsche Forschungsgemeinschaft (German Science Foundation) via grant number FOR 384 is gratefully acknowledged.
References: [1]
F. Mielentz, M. Krause, H. Wüstenberg, H. Wiggenhauser: Development of a phased array transmitting equipment for ultrasonic testing of concrete. In DGZfP; Proceedings of the 11th
International Symposium on Nondestructive Characterisation of Materials, 24-28.06.2002, Berlin,
DGZfP (2002) [5] F.Schubert: Propagation Characteristics of Ultrasonic Waves in Concrete and Conclusions for |
| © NDT.net |