International Symposium (NDT-CE 2003)Non-Destructive Testing in Civil Engineering 2003
|Start > Contributions >Lectures > Ultrasonic 1:|
ULTRASONIC IMAGING OF CONCRETE ELEMENTS: STATE OF THE ART USING 2D SYNTHETIC APERTUREMartin Krause, Frank Mielentz, Boris Milmann, Doreen Streicher
Bundesanstalt für Materialforschung und -prüfung, Berlin (Germany)
Fraunhofer Institut für zerstörungsfreie Prüfverfahren, Saarbrücken (Germany)
Since several years ultrasonic imaging techniques are used for testing concrete elements. Combined with a reconstruction calculation they lead to an acoustic imaging of scatterers and reflectors from the inside of the concrete element. For several testing problems it is an advantage to use a two dimensional synthetic aperture and to combine it with a three-dimensional reconstruction calculation (3D-SAFT, Synthetic Aperture Focussing Technique).
This enables the analysis of the tested volume by means of B-scans in two directions and C-scans (projections parallel to the surface), respectively. Using the 2D-scanning technique, especially the disturbing influence of the reinforcing rebars can be minimised.
In this contribution the state of the art of this method is demonstrated and analysed. Examples of the application are presented, concerning
The results presented are obtained in the frame of a research initiative Non-destructive Evaluation of Concrete Structures Using Acoustic and Electromagnetic Echo-Methods supported by the German Science Council (FOR 384 of DFG, Deutsche Forschungsgemeinschaft). The methods are assessed and improved in cooperation with the partners participating in the project.
During the last decade the ultrasonic echo technique has become an important tool for testing concrete elements. The applied ultrasonic frequencies are in the range from 40 to 200 kHz, corresponding to a wavelength of pressure waves in concrete from about 100 mm to 20 mm. The reason to do so is the inhomogeneous nature of concrete caused by the aggregates and the pores. Using long wavelengths, the pulse-echo technique is effective for the detection of internal objects, but it leads also to a strong backscatter of the incident pulse. This structural noise can mask the signals from the internal objects to be detected. In most cases the desired echo signal cannot be read from a single A-scan (reflected amplitude vs. time).
The ultrasonic echo methods achieved an enormous progress during the last years. This is due to the development of broadband transducers in the desired frequency range having a short pulse response, and the application of synthetic aperture measuring and evaluation techniques [1-3]. The reconstruction by Synthetic Aperture Focusing Technique (SAFT) offers a powerful approach in order to create high resolution images from the inside of the concrete element or specimen under inspection [4, 5].
This paper focuses on the application of the 3 dimensional SAFT reconstruction from two-dimensional measurements collected at many different points at the surface offering the possibility to evaluate B-scan images (intensity vs. depth along an axis) in two directions of projections and C-scan images (reflection intensity parallel to the surface in a chosen depth). The different experimental approaches are summarised in chapter 2, whereas the principle of the reconstruction is briefly described in chapter3.
The development and optimisation of testing methods consists of three important steps:
Point c) is subject of an intense co-operation with the working group of Prof. Langenberg (University of Kassel) . The modelling method used (Elastodynamic Finit Integration Technique, EFIT) offers the possibility to take into account the microscopic structure of concrete (aggregates, pores, grouting cement) and the structural elements (tendon ducts, tendons, multilayered systems). All wave modes and the conversion between them can be considered.
Another important testing method working with elastic waves is the impact-echo method. The comparison and combination of experimental and modelling results is an important subject of the research group FOR 384 funded by the German research council (DFG) .
In the present paper the achieved capacity of three-dimensional imaging of concrete elements is described for several testing problems:
There are several possibilities to collect data using a two dimensional aperture. Principally they are classified in monostatic, bistatic and multistatic methods .
The classical testing method is monostatic, applied e.g. for high frequency ultrasonic of metallic objects. For concrete, this method is usually applied with two neighboured transducers, moving over the surface in a meander line. The most frequently used coupling agents are vaseline or glycerine, the coupling has to be controlled at each point, because of difficulties with the possibly rough surfaces.
The bistatic method means that two transducers are moved over the surface having a fixed distance. This method fits best, when the echoes can be read directly from the A-scans. This method is only applicable for concrete elements, when the inspection conditions are very easy (e.g. thickness measurement without reinforcement).
For the multistatic approach there are several possibilities following the different needs.
For scanning large area surfaces, an array of ten transducers is used, for most applications put together in a template as shown in figure 1 [9, 5].The transducers have a diameter 50 mm, the bandwidth is in the range of 50 kHz to 240 kHz. For signal excitation an arbitrary function generator is used. In every position, the ten transducers act as transmitter and receiver one after the other governed by an electronic switch and a portable PC. In this way 45 pairs of transmitting receiving positions - that means 90 different A-scans - are collected in a relatively short time. For measuring a surface, the template is moved over the surface. The measuring time per point is typically two minutes; the usual step width is between 10 mm and 30 mm. The coupling has to be controlled at each position.
In order to reduce the problem of positioning and coupling transducers, test equipment using a scanning laser Doppler vibrometer with automated data recording was developed [10, 11, 12]. For these measurements a broadband transducer with a centre frequency of 100 kHz is used as transmitter. For reception, the laser vibrometer measures the surface movement in direction to the laser beam in the frequency range up to 250 kHz. In order to enhance the signal to noise ratio, the measured area is covered with auto reflecting foil or colour. For collecting the data, a time averaging system is used. The measuring point is moved over the surface using a mirror scanner as shown in figure 2; a typical aperture size is 200 mm x 200 mm, typically scanned on a 10 mm x 10 mm grid in 40 minutes.
A new developed measuring system is commercially available. It works with point contact transducers emitting shear waves with a centre frequency of 50 kHz (Flaw detector A1220) [13, 14]. The system consists of a sending/receiving unit of 24 point transducers and a small portable control unit. The transmit-receive unit works without coupling agent, it is used for scanning concrete surfaces. For the results described in this paper, the same pulse excitation and receiving unit as for the ultrasonic array is used. It has been shown that programmed pulse forms are well suited to achieve a good imaging quality and to suppress disturbing surface wave effects. Figure3 shows the application of the system at a concrete specimen at BAM containing a tendon duct (experiments described below).
Applying shear waves, the direction of the polarisation can be chosen. There are first results to use the different polarisation for the interpretation of defects in concrete, but the possibilities are not yet studied systematically.
In addition a pair of point contact transducers can be used in the multistatic mode. They are available as single transducers for shear as well as pressure waves. Since they don't need any coupling agent, they can be moved easily from point to point. There are first experiences available for crack characterisation in this measuring mode.
As mentioned above, there are several types of very capable broadband transducers available now in the desired frequency range. They can be driven by means of rectangular pulses, controlled pulses (RC2) or pulse sequences (e.g. chirps). The chosen pulse form depends on the application [15, 16]. Briefly summarised the rectangular pulses are well suited for time of flight measurement, because of the sharp leading edge. Adapted programmed pulses are optimal for multistatic measuring techniques, because they permit the best suppression of disturbing surface wave effects. They are used for the results described in the present paper.
Figure 4 shows the principle: the probe transmits ultrasonic waves into the concrete material, which are reflected, e.g. on a tendon duct and received inside a receiving aperture consisting of measurement points of a scanning laser vibrometer or a receiving probe-array. The received RF-signals are stored in a PC. For the different points inside the aperture the sound-paths and thus time-of-flights are different. During the reconstruction process these sound paths are calculated for each possible path transmitter-reflector-receiver and the stored RF-signals are superposed time-of-flight corrected leading to high reconstructed amplitudes at those points of the reconstructed image were reflection occurred and low amplitudes anywhere else. Thus the reconstruction process results in a 3D representation of the backscattered intensity from inside the specimen. To interprete such data fields, projections are plotted, which are equivalent to the B-scan (ultrasonic intensity along the scan-axis vs. depth), C-scan (ultrasonic intensity parallel to the inspection surface in an adjustable depth range), and D-scan images (ultrasonic intensity along the index-axis vs. depth). The computer time required for a 3D-SAFT reconstruction depends on the number of A-scan in the data set and the reconstruction depth. For the investigation of reinforcement bars in a concrete plate up to 9000 A-scans were collected. The reconstruction time on a 1.8 GHz PC was about 30 min.
Examples of application at specimens
In the following, the capacity of three dimensional imaging is described for three important testing problems for concrete.
Localisation of compaction faults and honeycombing
For measurements the transducer template described above was moved over the surface along a line with a step width of 20 mm. The results already described in previous publications showed that several of the styrodur balls and gravel pockets can be imaged with a deviation of only 10 mm corresponding to the construction plan [18, 5]. Also the lower layer of reinforcement can be imaged. Thus it is possible to estimate its concrete cover. Normally this is a task for impulse radar, the recommended method for localising all types of metallic construction elements, but which may be difficult for young concrete due to the higher moisture content. One advantage of the three dimensional imaging is to produce B-can images in other directions and C-scan images. In this case the B-scan images parallel to the Y-axis with the indicated projection segments confirm the interpretation of the results.
Characterisation of cracks
Crack characterisation was developed by imaging the forward scattering of pressure waves. The measurement principle corresponds to the multistatic approach. A scanning laser Doppler vibrometer is used as ultrasonic receiver as described above (figure 2). Several sending positions are combined to one data field, which is processed by means of reconstruction calculation (3D-SAFT) as described above. Thus the crack region can be imaged with B-scans parallel and perpendicular to the crack direction.
This method leads to a clear image of the crack tip as well as from inclined notches and differs between the scatter from contact bridges (e.g. unbroken aggregates and the crack tip.
Also real cracks can be imaged. The changing depth of the cracks is read from the corresponding B-scan image. The experiments indicate that a rather broad region of the crack above the crack tip scatters ultrasonic waves, certainly due to small sound bridges .
Investigation of tendon ducts
The first two inspection problems are solvable by means of impulse radar in most cases. But when the non-prestressed reinforcing rebars are dense, the electromagnetic shielding is too strong for applying successfully impulse radar. Applying ultrasound imaging, it has been shown that tendon ducts having a diameter of 85 mm can be imaged behind twofold reinforcement rebars (mesh width 75 mm, diameter 12 mm) . For this a scanning laser Doppler vibrometer was used.
Up to now much research activities were investigated in order to find out, how and under what conditions grouting defects in tendon ducts can be localised. Two possibilities of evaluation were found to indicate the difference between well and bad grouted areas of tendon ducts:
Parts of the tests were performed at specimen with grouted areas and artificial voids made with polystorol. Those specimens did not contain tendons for simplifying the problem and for investigating the basic principle of the method. Two test specimen containing tendon ducts with several voids were investigated during two round robin tests (one initiated by BAM , one funded by the Federal Highway Research Centre (BASt) [21, 19]). It was shown that artificial voids can be localised by scanning a linear aperture and L-SAFT reconstruction as well as 2D-scanning using a laser Doppler vibrometer and 3D-SAFT reconstruction. In most cases the difference in the reflection intensity following the principle a) indicates the ungrouted areas. On the other hand not all artificial voids could be found. The difference in the reflectivity is sometimes smaller than 6 dB, and so it is difficult to distinguish it from a change of the coupling conditions.
The test series at BASt were carried out as blind test. One of the voids was identified by the disappearance of the back side reflection of the duct (principle b). This was achieved by means of the 3D-SAFT reconstructions from a superposition of several areas scanned with a laser Doppler vibrometer. For the reconstruction a special integration mode was used.
Since 2001, when the activities for the Research group FOR 384 started, the ultrasonic scanning methods were systematically compared at the specimen used for the first round robin test (BAMPK1). A borehole having the same diameter (85 mm) and depth as the tendon duct was placed in order to investigate the same geometry with and without the thin duct sheet (0.5 mm). Quantitative parameters are defined for a comparison of the quality of the different imaging approaches .
A second test series was carried out at similar specimens at TU Darmstadt applying all concerned imaging techniques. One of the results was that the border between a grouted and ungrouted part of the duct is equally localised, when it is measured by 2D imaging (Linear Synthetic Aperture Focusing Technique; LSAFT), by 3D imaging (this article) and using point contact shear wave transducers (with commercial software A1220). On the other hand the results did not indicate large well grouted regions in the same duct. Here small delaminations caused by the shrinking of the grouting mortar are supposed.
Also transversal tendon ducts having normally a smaller diameter (40 mm) were successfully investigated by 3D imaging. The moving transducer array described above was used at specimens and during an application at a post tensioned concrete bridge deck. Comparing B-scan and C-scan images from 3D-SAFT reconstructions, an overlap of two duct end containing air inclusions could be localised .
The discussion of the numerous results show that the individual realisation of the interface between concrete, steel sheet and grouting mortar has larger influence in the sound propagation as expected. Aspects to be taken into account are
It was assumed that the way of producing the ungrouted regions in the ducts has an influence on the appearance of the interface zone. Thus BAM produced two specimens containing ducts with slightly abraded surfaces to realise a good bonding between the concrete and the steel sheet of the duct. The ungrouted regions were produced implementing a plastic tube. The tendons pass the tube, which was sealed at the ends. Only a small opening lets pass the grouting mortar from one side to the other. So a professional pressure method could be used for grouting the duct, as is usual for real prestressed constructions.
Figure 5 shows the construction plan of the specimen. In addition to the duct styrodur balls are placed, symbolising compaction faults in the concrete. The position of the plastic tube inside the duct symbolising the void was localised applying g-radiography and is indicated.
The investigation of the specimens applying the different ultrasonic scanning methods is not yet terminated. Up to now the best results of ultrasonic imaging were obtained applying the shear wave transmitting/receiving unit. It was controlled using the electronic system for the transducer array as described above (figure 1). The transmit-receive unit was moved over the surface with a step width of 2 cm.
Figure 6 shows the B-scan image from the 3D-SAFT reconstruction measured with a polarisation parallel to the tendon duct. The duct having a concrete cover of 260 mm is clearly imaged. Between x = 800 mm and 1400 mm the reflected intensity is enhanced by the factor 1.7, corresponding to the location of the void. But the difference in intensity would probably be not large enough, to localise it reliably in a blind test. Shadowing from the reinforcing rebars probably causes the pointed structure of the upper side of the duct (between x = 1000 mm to 1800 mm).
In the depth of 380 mm there is a second reflecting line only apparent in the grouted area of the duct. This can be interpreted as the reflection from the back side of the duct. The fact that the indicated depth is slightly larger than the depth of the backside, is not yet explained. The interpretation as back side reflection is confirmed by the different intensity at the left and the right side of the void in the same B-scan image. The higher intensity at the left side is caused by an unintended small grouting error (air inclusion), which is visible from the side surface shown in figure 6.
The interpretation is also confirmed by the C-scan images shown in figure 7. The intense reflecting area from x= 300 to 700 mm (in the depth of the back side) is 40 mm wide, corresponding to the breadth of the air inclusion. The imaged part from the upper side of the duct is smaller, as it is typical for round objects.
This result is impressive, because it implies that the ultrasonic pulses pass the duct including the tendons two times. Even if the result has to be verified later e.g. by modelling and destructive testing, it seems to give a clear possibility to image well grouted regions in observing the appearance of the back side reflection from the duct.
The back side reflection at ducts was already observed in the blind test at the specimen at BASt as described above  and during the investigation of a bridge girder. But since these experiments were carried out using pressure waves, the reflections from the back side are not clearly separated from signals resulting from mode conversion effects.
Ultrasonic 2D scanning methods in combination with 3D reconstruction by means of 3D-SAFT (Synthetic Aperture Focusing Technique) are a powerful tool for investigating concrete elements. In comparison to linear scanning and reconstruction they offer the possibility to visualise reflection and scatter versus depth in all desired directions (B-scan images), and parallel to the surface in a chosen depth (C-scan image). Thus the interpretation of the data can be more reliable than regarding only one projection. Different scanning systems are in use: Array of broad band transducers being moved over the surface, neighboured transmitting/receiving transducers, and scanning laser vibrometer as receiver. New developed point contacts transducers enable measuring without coupling agent. The different possibilities of measuring bistatic and multistatic are analysed and compared for different testing problems. The comparison of different imaging techniques is part of a co-operation in a research group of the German Research Council (FOR 384 of DFG).
It has been demonstrated at foundation plates containing artificial defects, that the described imaging technique is able to localise compaction faults and honeycombing behind rather dense non prestressed reinforcement. Also the concrete cover of the lower reinforcement near the back wall could be measured.
The imaging of cracks including measuring the crack depth is possible by measuring the forward scatter of ultrasonic waves at the crack tip. Here the scanning laser Doppler vibrometer acts as ultrasonic receiver. Since real cracks often contain sound bridges like unbroken aggregates and reinforcing, the crack tip can be identified as the lowest ultrasonic scatter. In order to indicate cracks going through the concrete element, the back wall echo is analysed.
Grouting faults in tendon ducts were investigated with pressure-waves and shear waves. Applying shear waves, it was clearly demonstrated for the first time that the back side of a tendon duct containing tending wires is clearly imaged in case of good grouting. Since this it not the case, when there are air inclusions or voids in the duct, a qualitative parameter for indication of grouting defects is demonstrated. Analysing the intensity of the direct reflection intensity of the ducts, the localisation of artificial voids in tendon ducts by means of ultrasonic 3D imaging is feasible, but not yet reliable. Further research has to be done in order to analyse the influence of the conditions of the boundary layer and the air pores.
Support on basic research is gratefully acknowledged by Deutsche Forschungsgemeinschaft (DFG, German Research Council) via grant number FOR 384. The specimens for the new tendon duct investigations were planed by Ch. Kohl and realised by the concrete laboratory of BAM VII.1. Dywidag Systems International grouted the ducts (DSI, Dipl.-Ing. H. Iven). BAM VIII.4 verified the location of the voids in the ducts using g-radiography. All support is gratefully acknowledged.