Bundesanstalt für Materialforschung und -prüfung

International Symposium (NDT-CE 2003)

Non-Destructive Testing in Civil Engineering 2003
Start > Contributions >Lectures > Ultrasonic 1: Print

Study of phased array techniques for cracks characterization in concrete structures.

O. Paris, Ph. Brédif, O. Roy, Commissariat à l'Energie Atomique, CEA/DECS, Saclay France
J. M. Rambach, G. Nahas, Institut de Radioprotection et de Sureté Nucléaire, IRSN, France


The concrete containment building of nuclear power plant (NPP) structures are susceptible to ageing by various processes depending on the operating environment and service conditions. In order to ensure the structural capacity and leak tight integrity of concrete structures, it is necessary to detect cracking phenomena which is one of the primary sign of ageing. The aim of the present work is to propose a NDT method able to characterize and to represent the crack shape inside the concrete wall. Ultrasonics were identified as having a great potential to detect cracks in such thick concrete structures. Typical frequency range used for ultrasonic control in concrete is low between 50 and 250kHz. This study drew its inspiration from ultrasonic process applied on steel components. We assume that a crack can be seen as a set of facets with different orientations. Each facet is considered as a reflector. Since the maximum size of the aggregates is 20 mm, we use transducers working at 250 kHz with large bandwith to detect reflectors with a minimum diameter of 20 mm. Such high frequency produces poor signal to noise ratio (SNR), so we use a combination of several transducers to increase the SNR. This paper deals with the results obtained with two different techniques applied to characterize the artificial crack : SAFT processing and Phased array techniques. All results are presented on reconstructed images taking account of the material characteristics and the ultrasonic beam orientation.

1. Introduction

Wall containments in Pressurized Water Reactor (PWR) are extremely important because they must retain radioactive materials if an accident should occur. These walls are thick (1.2 m or 0.9m) pre-stressed and reinforced concrete structures containing pretensionned cables. The containment is cylindrical with an hemispherical dome.

For the safety authority (IRSN) it is essential to maintain the integrity of the wall containment. This means to ensure the structural capacity and leak tightness of the concrete structure.

Potential degradations of the concrete material can affect the ability of a Nuclear Power Plant containment to perform satisfactorily its functions. The degradation of concrete can be the result of various attacks [1]. Physical attack mechanisms include freeze/thaw cycling, thermal expansion, thermal cycling abrasion. Chemical attacks may also occur in several forms : efflorescence, attack by sulphate and by acids, salt crystallisation and alkali-aggregates reaction.

Such mechanisms and fatigue effects can be responsible for surfaces and bulk flaws such as cracks, porosity and laminations.

The aim of our work, supported by IRSN, is to detect and determine the crack size inside the wall and to check the defect growth between two inspections. Current NDT techniques [2] such as visual inspections, radiography, impact echo allow to detect cracks in concrete under favourable conditions but not yet to characterize them. Ultrasonic is chosen as the best identified technique able to detect non destructively cracks in thick concrete structures.

Compared to inspection of steel components the difficulty of ultrasonic control in concrete comes from the heterogeneous structure. Concrete is a composite material constituted of a binding cement medium with embedded particles from sand to gravel with size up to 20mm. The propagation of a the wave in such media generates heavy scattering and attenuation of the sound energy that provide a resultant poor signal to noise ratio (SNR) of the reflected signal amplitudes.

For that reason typical frequency range used for ultrasonic control in concrete is quite low between 50 and 250kHz. In order to increase the SNR we propose ultrasonic techniques inspired from SAFT processing and phased array methods using a combination of several transducers.

Mock up calibration blocks were made in order to develop test procedures. One sample includes a known artificial crack composed of different orientated facets localised at different depth.

2. Material and Sample

The formulation of concrete used for that study is well representative to the one used in wall containment with a maximum size of aggregates of 20 mm. The concrete is composed of a cement medium with embedded particles of three repartitions of size, sand (0 to 5mm), gravel (8 to 2) and bigger gravel (12/20mm). The fraction of aggregates in total volume of concrete is around 2/3.

Fig 1: a) Sample of concrete b) 3D view of that sample c) Crack with the different facets "Pi" to be detected.

We carry out measurements on a concrete block of 600mm thick with dimensions of 800x800x600 mm3. We have introduced in that sample an artificial crack composed of four facets with different orientations. Each facet noted Pi on figure 1 has a surface of 40x70mm2 and is considered as a reflector for sound wave. These facets are supposed to represent the surface extension of crack, they are located between 200mm and 500mm below the surface of inspection. We assume that crack with an aperture of 100mm act as a perfect reflector for ultrasonic waves propagating in concrete.

Figure 1 shows the sample with the embedded artificial crack and flat holes. The flat holes have a square section of 50mm and have been used for reference measurements.

3. Instrumentation

The pulse echo technique is commonly used to detect and characterize cracks in steel components. Such ultrasonic inspection is based on the analysis of the backscattered signal from the inside of the concrete which is collected during the acquisition.

Fig 2: Experimental set-up used for the control of the concrete block.

As we try to detect with an increased spatial resolution small reflectors in a medium with aggregates of 20 mm, we use broadband transducers (diameter 38mm) operating at 250 kHz. In order to improve the signal to noise we use a set of six probes to perform signal reconstruction. Transducers are mechanically displaced in order to scan the surface of the concrete.

A multi channel system is used which allows to drive the 6 transducers, to adjust their relative parameters (delay at transmitter or receivers, amplitude..). The signal of excitation is 250V square pulse. The coupling is a film of water and the probes are spring loaded in order to compensate the variations at the surface of the block due to possible roughness.

4. Multi-probes reconstruction

Fig 3: Procedure of reconstruction.

In order to improve the signal to noise ratio of the measurements we apply the principle of reconstruction used in SAFT processing [3] with an arrangement of 6 broadband transducers. The set of transducers is displaced mechanically over the surface of the concrete to be scanned. One probe is used as an transmitter while others are functioning as receivers. The transmitter is then permutated in order to increase the number of acquisitions. After the acquisition, the reconstruction consists in a combination of the recorded signal for each channel and position in scan axis after a fit in time and position (figure 3). Such a fit takes into account the delay of the reflected wave introduced by the transmitter-receiver distance. This reconstruction must have a constructive effect on the echo of flaw and a destructive one on the noise which leads to an improvement of the signal to noise ratio.

We apply this procedure to detect the different facets of the artificial crack. The reconstructions are made on Bscan images which represents the signal along the scan axis versus depth. The transducers are disposed as shown in figure 4a, the smallest distance probe to probe is 80mm. We present on figure 4b the result of reconstruction from 10 raw Bscans. A Bscan image corresponding to the signal recorded with a single transducer in pulse echo mode is also presented to make comparisons.

Fig 4: a) Arrangement of probes and acquisition b) Comparison between the mononelement measurement of the crack and the reconstructed Bscans with theoretical parameters.

The variations of colour indicate areas of different acoustical response which can corresponds to a region of flaw. We observe on results presented in figure 4 an important structural noise in mono element case which is reduced in the reconstructed case. The reconstruction is successful, all the facets to be detected are reconstructed together. This shows that the applied parameters in time and position are quite well determined. The reconstruction is efficient in many points : facets up to 500mm under the surface of inspection are detected with an improved SNR, the structural noise is importantly reduced compared to the monoelement case and there is a good match in time and position of the theoretical profile of the crack (in white in figure 4) with the detected echoes coming from the different facets.

5. Multi phased array measurements

We use the simple arrangement of five transducers described in figure 5 in order to test the ability to focus a ultrasonic beam in concrete structure like in steel components [4]. The aim is to enhance the energy in a selected depth in order to increase the resolution and signal to noise ratio. Such a symmetric "star" arrangement allows to test the focussing at different depths just by adjusting the delay of the central element.

Fig 5: "Star" arrangement of transducers in order to test phased array technique.

System characterisation
First we test experimentally the focussing on the back wall of the block at 600mm in order to characterize the transducer configuration. The measurement is done in transmission configuration trough 600mm as illustrated in figure 6 : the 5 transducers are used as transmitter and disposed on one surface of the block. Another transducer working as receiver on the opposite surface. We observe the resulting Cscan image on figure 5 which represents the maximum amplitude of the signal parallel to the surface in the whole inspected depth. The focal spot extends at -3dB over 170mm while the mono element measurement with a unique transmitter gives 250mm at -3dB. So we observe the expected reduction of the focal spot due to focalisation.

Fig 6:
Measurement in transmission configuration to test focussing of 5 transducers in star configuration compared to the mono element case.

Fig 7:
a) probes and acquisition b ) Cscan measured with focussing on hole flat at 200 mm depth.

Fig 8:
Bscan showing profile of crack by focussing of five transducers, comparison with the multi probes reconstruction.

Figure 7 shows tests of focussing on flat holes T1 and T4 localized at 200mm and at 400mm depth. By applying theoretical delays the holes are detected with good signal to noise ratio, 10 dB at 200 mm and 8 dB at 400mm.

We apply focussing with same configuration on the crack by adjusting delay law in order to detect each facet. The result is shown on figure 8 on Bscan image. All facets are detected but the SNR for the deeper ones is weak. We recall the multi probes reconstruction of the crack obtained in figure 4 in order to make comparison with the mono element case. Phased array technique is better than multi probes reconstruction for the detection of deep reflectors, for instance P5 shows a better SNR and a better lateral resolution. But close to the surface of inspection multi probes reconstruction is more efficient because we are not disturbed by the dead zone effect due to the aperture of the transducers array.


The aim of that study is to characterize non destructive cracks in concrete of nuclear power plant. We have presented results with two ultrasonic techniques applied to the detection of an artificial crack embedded in a concrete block.

Multi probes reconstruction using principle of SAFT processing have been tested from results of experiments performed with a set of several transducers. The reconstruction using theoretical parameters of reconstruction is effective to detect with an improved SNR the different facets composing the crack. Tests of focussing of the ultrasonic beam from five transducers have been also performed successfully at different depth for the detection of flat holes and crack. Phase tests show he great potential of phased array techniques applied to concrete structure inspection. Complete control of the array of probes at transmission and at reception will allow to improve the performance at various depths. Next test will concern the detection of tilted reflectors.


  1. "Review of NPP concrete degradation factors and assessment methods" ,T. M. Refai, Processing of structures congress, Texas, 1992.
  2. "Imaging of concrete structures" O. Buyukozturk .NDT& E Interbational Vol 31 1998
  3. "Towards SAFT imaging in ultrasonic inspection of concrete" M. Shickert, International Symposium Non Destructive Testing in Civil Engineering (NDT-CE) 411-148 September 1995
  4. "Improvement of defect characterization with ultrasonic adaptive focusing technique"
  5. S. Mahaut, G. Cattiaux, O. Roy and Ph. Benoist, 14th Int. Conf. on NDE in the Nucl. and Pres. Ves. Ind., p. 427 (1996).
STARTPublisher: DGfZPPrograming: NDT.net