Classification of cracking mode in concrete by acoustic emission parameters

https://doi.org/10.1016/j.mechrescom.2011.03.007Get rights and content

Abstract

The study occupies with acoustic emission monitoring of several types of concrete during bending. The signals emitted at the different fracturing stages exhibit distinct signatures. Specifically, frequency and shape parameters of the acquired waveforms shift during the experiment, closely following the sequence of fracture mechanisms from tensile micro-cracking to brittle macro-cracking and fiber pull out. A number of AE indices are proposed, the use of which will enable classification of the cracks according to their mode. The study sheds light to the fracture process of cementitious materials, and enables a warning against the final failure. The simplicity of the scheme renders it applicable in situ.

Introduction

The importance of monitoring of the structural safety of concrete structures has been long stated. Early assessment of material condition against large-scale failure helps to manage the structures safely and economically. One of the methods used for real time nondestructive monitoring is acoustic emission (AE). Using this technique, the elastic waves released during crack propagation incidences are recorded by transducers placed on the surface of the material (Grosse and Ohtsu, 2008, Mindess, 2004). The transducers are usually piezoelectric and transform the energy of the transient elastic wave to an electric waveform which is digitized and stored. The information of these waveforms includes the location of the source crack (by comparing the arrival time to different sensors), the density of cracks, as well as the severity of the materials condition (Kurz et al., 2006, Aggelis et al., 2010a, Carpinteri et al., 2010, Matikas, 2008, Zhou et al., 2010). A very important aspect is the correlation of AE indices to the cracking mode. For most cases of materials and loadings until failure, tensile cracks are developed at the initial stage of loading, while shear cracks dominate later (Yuyama et al., 1999). This is typical for reinforced concrete members that undergo bending. The initial cracking comes from the tensile load on the surface of concrete, while the member ultimately fails with diagonal shear cracks (Ohno and Ohtsu, 2010). Therefore, it is beneficial to characterize the mode of the cracks as it can lead to early assessment of the materials condition. For this goal, Moment Tensor Analysis (MTA) provides information taking into account the first cycle of each AE signal. The results have been demonstrated in laboratory conditions in various experiments successfully classifying the cracks depending on their mode (Ohno and Ohtsu, 2010, Kawasaki et al., 2010). However, in certain cases of structures the application is not straightforward mainly because of the number of necessary sensors to detect each cracking event (at least six). The crack location is not known a priori, and therefore the sensors are distributed evenly to cover the largest amount of material volume. Thus, the distance between neighboring sensors does not allow capturing one single cracking event by the necessary number of sensors. Especially in long structures like bridges, the sensors are placed in a straight line in order to cover as much as possible of the structure's dimension (Shiotani et al., 2009). This placement enables only linear location of the damage zones but is not suitable for MTA. Therefore, simple characterization schemes are sought for, which would enable characterization of the cracking mode based on the information of a small number of sensors.

It has been seen that a crack propagation incidence has different acoustic emission signatures depending on the mode of the crack. The tensile mode of crack which includes opposing movement of the crack sides, results in AE waveforms with short rise time and high frequency. On the contrary shear type of cracks usually result in longer waveforms, with lower frequency and longer rise time (Aggelis et al., 2010b). This is mainly due to the larger part of energy transmitted in the form of shear waves, which are slower; therefore, the maximum peak of the waveform delays considerably compared to the onset of the initial longitudinal arrivals. This has been demonstrated in different kinds of materials, like concrete (Ohno and Ohtsu, 2010, Soulioti et al., 2009), fiber composites (Philippidis et al., 1998, Aggelis et al., 2010c) and rock (Shiotani, 2006). Fig. 1 shows a typical AE signal after a crack propagation event with its main features. One of the crucial parameters which are influenced by the mode of crack according to the above discussion is the average frequency, AF which is defined by the ratio of threshold crossings over the duration of the signal and is measured in kHz. It is one estimate of the basic frequency content of the waveform. Another crucial parameter is the RA value which is the rise time (RT, delay between the onset and the maximum amplitude) over the amplitude, A, measured in μs/V. AE energy (ENE) is another important parameter expressing the measure of the area under the rectified signal envelope (MARSE). It is reasonable to suggest that the energy in the acoustic waveform is proportional to the energy of the associated deformation (Curtis, 1975) and is presented in dimensionless form (Philippidis et al., 1999). It expresses the severity of an event since large crack propagation incidences will emit larger amount of energy than micro-cracks. These parameters have been used to characterize the cracking mode in laboratory conditions in accelerated corrosion experiments in reinforced concrete (Ohtsu and Tomoda, 2007), in bending of concrete reinforced with metal bars (Ohno and Ohtsu, 2010), with steel fibers (Soulioti et al., 2009) as well as vinyl fibers (Aggelis et al., 2009). The results have shown that the emissions during the early damage stage (corresponding to tensile mode) exhibit higher AF and lower RA, while as the material is led to final failure AF decreases and RA increases. Based on above studies, it can be concluded that tensile matrix cracking results in AF higher than approximately 300 kHz and RA lower than 500 μs/V as measured by broadband transducers (Aggelis et al., 2010d). When sensors of low resonant frequency (nominally 150 kHz) are used the values 60 kHz and 2000 μs/V are suggested for AF and RA respectively (Soulioti et al., 2009). It would be desirable to test the above conclusions over a wide variety of concrete materials in order to estimate if the validity of any such criteria is global, or if they should be used in a case-specific basis, depending on the type of the material. This is of paramount importance in the field of engineering, while it is the subject of relevant scientific committees aiming at standardization of the AE testing in the concrete field (Ohtsu, 2010).

The shape of the AE signal apart from its origin is influenced by other conditions. One is scattering while propagating, which results from the inherent inhomogeneity of concrete due to its microstructure (sand, gravels, air voids, pores as well as cracking). Another influence comes from the differential velocity of the distinct wave modes excited by the source crack, as well as the mode conversion on the surface. This is inevitable since in AE testing the signal is acquired from the surface and is certain to include all different wave modes, including Rayleigh waves after impinging on the surface. Since the velocities of the modes are different, the waveform will be continuously spreading in time and the actual shape recorded will depend on the location of the sensor relatively to the crack. Additionally, the frequency characteristics of the sensors are crucial since they may mask the original frequency content of the wave. In this paper, the above factors are not thoroughly examined; the specimen size is quite small in order to exclude significant accumulation of scattering effects, as would occur in an actual size structure of several meters. On the other hand, the sensors are broadband in order to apply the least possible change in the content of the incoming signals. In any case the position of the sensors was fixed for all specimens and therefore, any change in the AE behavior during the fracture experiment is attributed to the successive fracture modes. In the present study, the acoustic emission results of a long series of fracture tests in different types of concrete are discussed. The different types include plain concrete and steel fiber reinforced concrete with different fiber contents, shapes and coatings, as well as water to cement ratio, making a population of approximately 100 specimens. They were tested in four-point bending with concurrent monitoring of their AE activity. Each of these distinct material characteristics, i.e. shape of the fibers or existence of chemical coating, are reflected to certain discrepancies on the recorded AE behavior (Aggelis et al., 2010d, Aggelis et al., 2010e). Nevertheless, the general behavior concerning the trends of AF and RA, as well as other AE parameters allows general conclusions for all materials independent of the mix and fiber parameters, based solely on the damage status; i.e. microcracking stage before the main crack formation, during the main fracture and afterwards as the damage is being continuously accumulated. The aim is to test the validity of a simple cracking mode characterization scheme based on specific AE parameters in laboratory. For this kind of AE parameter analysis a wide database is essential in order to increase the statistical importance of the conclusions. The present study includes approximately 100 specimens of different types of concrete which can be considered a significant population, compared to previous studies.

Section snippets

Experimental

Twenty five concrete mixtures were produced in total. Each mixture included four specimens. The specimens were prismatic, with dimensions 400 × 100 × 100 mm. The water to cement ratio by mass (w/c) varied from 0.5 to 0.6 and 0.7. The maximum aggregate size was 10 mm and the steel fiber contents varied from 0% (plain concrete) to 0.5%, 1%, 1.5% and 2%. Three different basic shapes of steel fibers were used, namely undulated (wavy), fibers with hooked ends, and straight. The thickness of the fibers

Conclusions

The present study discusses AE parameters to enable a simple classification scheme between different modes of fracture in concrete. A large population of specimens was fractured with concurrent monitoring of AE. Different AE parameters like the average frequency, RA and energy exhibit strong sensitivity to the fracture mode (tensile micro-cracking, macro-cracking or fiber pull out), showing that they can be included in a simple but reliable characterization scheme concerning the mode of damage

Acknowledgement

Dr. Dimitra Soulioti of the Laboratory of Concrete Technology and Non Destructive Evaluation of the Materials Science and Engineering Department of the University of Ioannina is gratefully acknowledged for the preparation of the specimens and the execution of the bending experiments.

References (26)

  • D.G. Aggelis et al.

    Advanced acoustic techniques for health monitoring of concrete structures

  • D.G. Aggelis et al.

    Acoustic emission and ultrasound for damage characterization of concrete elements

    ACI Mater. J.

    (2009)
  • D.G. Aggelis et al.

    Assessment of construction joint effect in full-scale concrete beams by acoustic emission activity

    J. Eng. Mech.

    (2010)
  • Cited by (600)

    View all citing articles on Scopus
    View full text