·Home ·Table of Contents ·Civil Engineering

Damage Level Evaluation and Characterisation by Acoustic Emission and Acousto-Ultrasonics in Concrete Under Compressive Loads Apostolos Tsimogiannis, Athanassios Anastassopoulos ENVIROCOUSTICS S.A. El. Venizelou 7 & Delfon, 11452 Athens, Greece E-mail: envac@otenet.gr Barbara Georgali HELLENIC CEMENT RESEARCH CENTER LTD. 15 K. Pateli Str., 14123 Athens, Greece Contact

ABSTRACT

The level of damage sustained by concrete structures is an area with increasing interest. At present monitoring damage in concrete is performed mainly by microscopic analysis which is a destructive method and requires a significant time interval to produce results and by simple compressive strength tests which also requires sample removal. Localized inspection by Non Destructive Testing (NDT) methods encounters a number of difficulties in concrete structures due to the nature of the material and can thus be applied only in a limited number of cases and usually when macro-cracking or other form of damage exists. Acoustic Emission (AE) has been used successfully in monitoring damage accumulation, crack growth etc. in large concrete structures when loads can be applied (bridges, piers etc.). Location of damage is possible along with other parameters (crack growth, orientation etc.). In structures where load application is difficult (e.g. buildings) the method is not used. To monitor damage on buildings Acousto-Ultrasonic (AU) is suggested as a technique. AU has been used to quantify macro-cracking in concrete structures. In the present work mortar cement specimens are being monitored by AE and AU during compressive tests to failure. AE is used to confirm microscopic damage in the specimens under loading. The state of the concrete is simultaneously monitored by AU and an attempt to correlate damage level as given by AU with AE measurements is made. The attempt is to characterize damage at early stages and produce an indication of approaching failure of the material. An attempt is also made to correlate AU/AE parameters to the mechanisms producing the microscopic damage in the specimens at early stages using microscopy findings for various times in the specimens' lives. The results show clear indications that AU can detect sustained damage at the microscopic level in concrete and can give early indications of oncoming failure. In addition AU provides a means to monitor and characterize events at the microscopic level of concrete in real-time.

1 INTRODUCTION

The current work was initiated following a severe earthquake in Attika, Greece of magnitude 5.9, with the epicenter 10km from the center of Athens in September 1999. The loss of life and damage caused was severe and the need for a means to assess the condition of buildings in general was emphasized. It has also been noticed that some of the collapsed buildings had sustained damage from previous earthquakes, the interaction with the atmosphere and through aging which resulted in a weakened structure that could not withstand this event. In addition, the question was raised about the effect of this event to buildings which, at the time, were in the construction phase and the concrete had not yet reached a significant strength level. For this reason experiments were made at 2, 7 and 28 days of the specimens lives. Quantification of the damage caused on a concrete based structure is very difficult to achieve and destructive or, in general, intrusive methods are required. The structure of concrete is such that it prevents use of traditional NDT methods to assess its internal condition. In addition damage in concrete can come from various sources such as environmental (chemical) and loading (e.g. earthquakes). The evaluation of the condition of a building that is suspect of defects and weakened frame is based largely on experience and, mainly, other than NDT methods. When severe damage is present, such as macro-cracks and visible surface cracks, experience is used to assess the condition of the structure. If these are not present but the structure must be investigated for integrity semi-destructive methods are usually used. It is therefore of importance to characterize damage and assess its criticality and the level at which the concrete has been weakened by faster means and non-destructively. This, certainly, involves more information than a single NDT method can supply. An NDT method can supply some information about the internal condition of the concrete. Therefore it is seen as important to fuse data from more than one NDT method in an attempt to evaluate and characterize the level of damage in concrete. In addition such methods, acquiring information during the loading process of, initially, mortar specimens, can supply important information about the behavior and the processes in the microstructure of concrete.
Acoustic Emission (AE) has been used with much success in monitoring large concrete structures such as bridges, dams etc. where a load can be applied and varied so as to excite defects and acquire information regarding their position, size, mode etc. leading to an assessment of the "threat" a defect poses to structural integrity. The method covers the complete structure with an adequate number of sensors but requires loading, which is very difficult to achieve for buildings.
Acousto-Ultrasonics (AU) uses the same hardware technology as AE but relies on sending a signal which will interact with the material on its path to the receiver. The changes introduced by this interaction make the signal a carrier of information regarding the structure of the material on its path. Because of the lower frequency than traditional UT it can penetrate through more concrete and still have the resolution to be affected by very small changes.
Analysis and discrimination of the fused AE/AU data was possible through the application of Pattern Recognition (PR) methods, which enhances the ability to separate and monitor the data through all specimens.
A combination of the two techniques gives the advantage of being in a position to monitor in an experimental phase damage by AE during compression and to correlate with AU which requires no loading.

2 MATERIALS & INSTRUMENTATION

At the first stage of experiments mortar specimens were used. The mortar cement mixtures used for the present work are presented in Table 1 with some details regarding their composition. For each type and age two specimens were tested. The specimens were kept in a 90% humidity controlled temperature environment as required by standards for concrete and were removed from this environment a few hours (2-3) prior to testing. A brief discussion on two concrete specimens, which initiate the second experiment phase, is given in section 6. The data from these two specimens are used for illustration only and are not included in the overall analysis as their composition is different. The two concrete specimens have W/C=0.525 and A/C=4.

ID	Age (days)	A/C	W/C
A	2	4	0.60
A	7	4	0.60
A	28	4	0.60
B	2	3	0.50
B	7	3	0.50
C	2	3	0.65
C	7	3	0.65
Table 1: Specimen specifications and data.

The equipment used were a PAC MISTRAS-2001 AE system and a PAC C-101-HV pulser. The sensors used for pulsing and receiving were PAC R6, 60kHz resonant with PAC 1220A preamplifiers. The sensors used were chosen for their high sensitivity, frequency response for concrete measurements[6] and their ability to be used for ultrasonic velocity measurements according to ASNT recommendations[2].
Microscopy results are available for specimens A for all three ages. Prior to the test each specimen was loaded to 150kN to remove AE produced from debris (dust, small particles etc.) on the surfaces of the specimen and to allow for any irregularities on its surface to settle. All specimens were compressed to failure and monitored throughout simultaneously by AE and AU. During compression one AE sensor was constantly receiving AE signals from the specimen. The AU pulses were generated at specific loads. Synchronous triggering of pulser and AE systems was used. The AE acquisition system recorded both features and waveforms for further analysis.

3 MICROSCOPY RESULTS

The results from microscopy were later used in an attempt to correlate AE data with findings at the microscopic level such as air voids etc.

Fig 1: Microscopy results for specimen A, all ages.


Microscopy is the main method used for the investigation of concrete and mortar microstructure. The application of optical microscopy to building materials such as concrete and mortar investigates the quality of aggregates, cement paste characteristics, water to cement ratio (W/C), air voids, porosity, cracking etc. For the purposes of the present work air void characteristics were established for all specimen ages according to ASTM C-457[8]. The microscopy inspection is performed on thin samples (<25mm) at magnifications 30x - 250x. The method is based on statistical considerations regarding the measured parameters and was performed manually as opposed to using an image analyzer software. This may introduce an error which is operator dependent. The current work was performed, for consistency by a single operator. Typical microscopy results are shown in Figure 1 for specimen type A for all ages. The presented parameters are:
Air Void Content : The included air in the specimens as a percentage of the total volume. This parameter remains practically constant. The small variations are of no significance to the structure of the specimen.
Cement Paste Content : The percent of cement paste to the total volume. This parameter tends to increase with age and it is affected by the aggregate to cement ratio (A/C). Specimens with lower A/C have higher cement paste content.
Specific Surface : This parameter gives a measure of the relative size of air voids in the material. It is expressed as the ratio of the surface to the volume of these voids[8]. Its tendency is to increase with age and along with the fact that the overall void volume remains relatively constant it can be said that the average void size is decreasing with age.
Spacing factor : This parameter expresses the density at which voids appear in the material and is expressed as the maximum distance from any point in the cement paste to a void. This parameter decreases with specimen age.

4 ACOUSTIC EMISSION

4.1 AE Data Analysis Philosophy and Pattern Recognition
Fusing data from AE and AU (acquired simultaneously) required separation at the analysis stage. A usual approach in AE analysis is the treatment of data as a whole and the observation of characteristics such as signal signature, cumulative energy and its evolution etc.
Due to the complexity of investigating AE data for all specimens and attempting to correlate with other findings it is difficult for the analysts to draw conclusions and much effort is required. The conclusions presented in section 4.2, although significant, contain only a small fraction of the information contained in the AE data as they rely on overall observations. The application of Pattern Recognition (PR)[9][10] to the data was made in an attempt to overcome such problems and limitations. The method is numerical so care must be taken for the preparation of the data. Due to the large dynamic range in some of the AE features a non-linear axis conversion was performed based on logarithmic functions. Subsequently the data were normalized so as to achieve a non-dimensional space. Envirocoustics SA pattern recognition software (Noesis)[7] was used for all PR and data manipulation. The Max-Min Distance[7][9] algorithm was applied to the data for the discrimination of signal groups and the set-up of a starting point for the analysis. Based on the results a k-NNC[7][10] supervised PR algorithm was trained to an accuracy greater than 97.8%. This algorithm was then applied to all sets. Typical results are shown in Figure 4a and 4b. The success of the algorithm to distinguish the AU pulser and receiver signals from other data. At its final form the algorithm had 100% success in recognizing pulser signals. Having such a means to separate the data the progress of different types of signals, which may signify different signal source, can be monitored and more detailed conclusions can be made. The following sections describe the results from these procedures.

4.2 Overall AE Activity
The data collected during compressive loading to fracture of the specimens provided information regarding the initiation and evolution of changes in the material and resulting damage.
The results were such, so as to provide indications about differences in the manner in which the specimens fracture and possibly the mechanisms that cause damage depending both on age and composition of the specimens. This is observed in Figures 2 and 3 were various overall activity data are presented. Figure 2 presents a comparison between specimens of the same composition for three ages (2, 7 and 28 days) for overall AE activity (number of events).

Fig 2: Total number of AE signals for all specimen ages.

Fig 3: Total number of signals for all specimens age 2 days.

As can be seen the changes in the overall behavior are distinct as are the slopes of the curves with younger specimens giving off more emission at lower loads. The 7 day specimens for all compositions gave less emission than any other age (see Figure 2b) and the 2 day specimens produced more medium energy emission at lower loads which diminished as the load was increased (see Figure 2a and 3a,b,c). Specimens at the age of 28 days had a distinct behavior with relatively little emission until the fracture load was approached. The overall activity, though, was not significantly higher than the 2 day specimens. Because the 2 and 28 day specimens have very different mechanical properties it can be said that the emission, to be more or less equal, has to come from different sources (mechanisms).
If the AE results are compared to the microscopy findings it can be observed that the lower emission at 7 days may be due to factors such as the small reduction in overall air void content. The overall activity seems to depend on the A/C ratio (see Figure 3), whereas the effect of the W/C ratio appears to be small. Figure 3 presents a comparison of the three compositions used at the same age (2 days). It is evident that specimens B and C (A/C=3) produce more AE at intermediate loads which reduces as the fracture load is approached, than specimen A (A/C=4). The changes introduced by variations in the W/C ratio have more of a qualitative effect. It is obvious, though, that the amount of information is such that it renders the task of correlating AE to microscopy difficult through classical AE analysis.

4.3 PR Class Analysis
PR analysis has provided the mentioned discrimination in AE signal classes (denoted C1-C6, Figures 4a,b) and AU signals (denoted Pulser, Receiver, Figures 4a,b). A brief description of conclusions made through the investigation of microscopy, AE and AU measurements for some of the classes is presented in Table 2.

Fig 4: Typical PR results. a) Energy vs Time, b) Signal Strength vs Absolute Energy.


In more detail, group C1 comprises low energy signals which appear at higher loads and increase in number as the specimens age. (See Figure 5a and 5b). A correlation to the A/C ratio could not be accurately established due to the number of available specimens. Nevertheless, the tendency is for this group of signals to reduce in size as the A/C ratio is reduced. This group of signals may be produced by mechanisms which affect the cement paste or produce friction in the cement paste at a very small scale. Group C2 contains signals with medium energy which appear at low loads. This group of signals appears at, mainly, intermediate loads and the amount of emission in this group diminishes with age. Due to this behavior it can be said that this group is produced by mechanisms relating to the aggregates and their bonding with the rest of the mixture. This is additionaly supported by the fact that lower A/C ratio specimens produced more emission in group C 2. Signals in groups C3 and C6 are high energy signals which appear at various loads and they tend to appear at increasing loads with age. These signals may be generated by disbonding of aggregates and macro-fracture as they exhibit some correlation to AU measurements. Group C6 is of particular interest as it contains only a small fraction of the total signals but about 30% of the total emitted energy which, in addition, remains constant for all ages. Thus group C6 may be related to actual macroscopic damage such as crack propagation.

Fig 5: a) Percent signals per class vs age, b) Percent energy per class vs age.

5 ACOUSTO-ULTRASONICS

The distinct feature of AU measurements is the fact that the method does not require load and produces a very different type of information than AE. Thus, a possible correlation between AE and AU readings will provide an NDT that can be applied to assess damage without the use of loading. The initial observation regarding AU measurements was that oncoming failure can be foreseen at about 15-20% before fracture load. The AU signal characteristics show a sharp, constant drop as the fracture load is approached.

Fig 6: Signal Energy Curve for AU signals and energy per class for AE signals.

Fig 7: Ultrasonic velocity diagram for all ages


During the experiments AU signal characteristics demonstrated significant variations even at low loads, being sensitive to changes in the material's microstructure (see Figure 6). Correlating AU to AE data provides further information about the ability of AU to indicate damage or changes, in general, in the material. AU measurements were also used to produce ultrasonic velocity results which is a standard NDT method for inspecting concrete structures. It is apparent from Figure 7 that although the velocity is affected by age it does not possess the sensitivity to indicate changes in the material at lower loads. Only as the load is raised to above 90% fracture load is there an indication of damage or oncoming failure. It is important to mention at this point that visual indications of damage appeared at greater than 95% fracture load. The use of extracted signal features (such as amplitude, energy etc.) from the received signal showed significant changes in all specimens and all ages with a distinct increase in received signal energy at intermediate loads. This is mainly caused by an increase in the material density and thus, the material may indeed become denser due to void collapse etc. as a result produce AE but without sustaining damage. Any sustained alteration to the material (such as macro-cracking) will appear in the AE data as well. When high energy classes appear AU measurements increase or decrease quite sharply (see Figure 6). A decrease below the initial energy level (prior to loading) will almost certainly indicate that damage has been introduced in the specimen. A sustained decrease indicates the approach to failure. Therefore, it may be possible to fully correlate AE and AU measurements and produce a means of non-destructively assessing the condition of concrete.

6 CONCLUSIONS

It has been demonstrated in the present work that a combination of NDT techniques that complement each other can produce important results for the understanding of the behavior of building materials under loading. Data fusion can provide the necessary tools to correlate measurements of the two techniques used, AE and AU, and produce an NDT technique for the assessment of the state of concrete in buildings. Furthermore it has been made obvious from the work involved that modern AE analysis can benefit drastically from the use of advanced analysis tools and pattern recognition.
Traditional techniques such as ultrasonic velocity measurements proved inadequate for detailed analysis. Experiments on real structures damaged by earthquakes indicated that the use of features as opposed to velocity produced a more sensitive means for crack detection and easily located cracks on columns which were almost invisible.
Initial experiments on concrete specimens, using the methodology outlined in this work (specimens Dc), show similar behavior to the mortar specimens. The presence of large aggregates produces more types of signals which will be analyzed in a similar manner.
In general due to the nature of building materials such as concrete, mortar cement etc. the use and development of new NDT techniques is necessary.

7 Reference

1. K. Matsuyama, T. Fujiwara, A. Ishibashi and M. Ohtsu : Field Application of Acoustic Emission for the Diagnosis of Structural Deterioration of Concrete, Journal of Acoustic Emission, vol.11, No.4, Oct-Dec 1993, pp.65-73.
2. ASNT, Non-Destructive Testing Handbook, vol.7: Ultrasonic Testing of Bridges and Buildings, pp 680-684.
3. S. Yuyama, T. Okamoto, M. Shigeishi, and M. Ohtsu : Quantitative Evaluation and Visualization of Cracking Process in Reinforced Concrete by a Moment Tensor Analysis of Acoustic Emission, Materials Evaluation, Vol.53, No.6, 1995, pp. 751-756.
4. S. Yuyama, T. Okamoto, T. Kamada, M. Ohtsu, T. Kishi : A Proposed Standard for Evaluating Structural Integrity of Reinforced Concrete Beams by Acoustic Emission, Progress in Acoustic emission VIII, The Japanese Society of NDI, 1996, pp. 295-304.
5. M. Uchida, T. Okamoto, T. Shibata, D. Mori, and M. Ohtsu : Structural Integrity Evaluation on Reinforced Concrete Members.
6. A. Tsimogiannis, A. Anastassopoulos : Experimental Set-up Design for Acousto-Ultrasonics on Concrete Specimens, MHKKYNES Project, Athens, 1999
7. Envirocoustics ABEE, Noesis Reference Manual, rev.0, 1999.
8. ASTM C 457-89, "Microscopical determination of air void content and parameters of air void system in hardened concrete"
9. A. A. Anastassopoulos, T. P. Philippidis, " Clustering Methodologies for the evaluation of AE from Composites ", J. of Acoustic Emission, Vol. 13, Noa 1/2, 1995, pp 11-21.
10. A. A. Anastassopoulos, V. N. Nikolaidis and T. P. Philippidis, "A Comparative study of Pattern Recognition Algorithms for Classification of Ultrasonic Signals", Neural Computing & Applications, 1999, Vol. 8, pp 53-61.

© AIPnD , created by NDT.net

|Home|    |Top|