LINEAR AND NON-LINEAR FREQUENCY DOMAIN TECHNIQUES FOR PROCESSING IMPACT ECHO SIGNALS TO EVALUATE DISTRIBUTED DAMAGE IN CONCRETE.

Abstract

The condition evaluation of in-situ concrete with non-destructive testing is difficult at best. The concrete deterioration processes of alkali-silica reaction (ASR), delayed ettringite formation (DEF) and freeze-thaw cycles all produce distributed damage in the form of micro-cracking which results in loss of strength or stiffness. Presently, a suitable field applicable method for determining the degree of micro-cracking does not exist. The impact echo test is potentially the best candidate if improvements can be made in the signal processing techniques which are crucial for accurately interpreting the data retrieved from concrete with distributed damage.

In this research, two batches of concrete specimens were prepared in accordance with standard procedures. A portion of each batch was subjected to either the Modified Duggan cycle or to Freeze Thaw cycles, both proven methods of inducing DEF and micro-cracking respectively. Curing techniques and materials were also chosen to accelerate distributed damage in the concrete specimens. In addition to the impact echo, a number of secondary tests were employed to monitor the progress of distributed damage in the concrete specimens.

Previous research efforts utilizing the impact echo method have attempted to characterize damage in terms of P-wave attenuation or pulse velocity. This involves signal processing in the time domain. These are inherently linear dynamics methods whereas the development of micro-cracks in concrete, an inhomogeneous material, gives rise to non-linear dynamics. Non-linear approaches to signal processing in the frequency domain are proposed herein. One involves calculating the deviation of the peak of the response spectrum from the shape of an ideal Lorentzian function model. The other calculates the second order non-linear harmonic coefficient.

The results showed that the potassium content, the curing methods and the Duggan and Freeze Thaw cycles had the desired effect of inducing distributed damage. The results of the signal processing of the impact echo data yielded more reasonable results for the specimens subjected to Freeze Thaw testing than for the specimens subjected to the Duggan Cycle. The results of the Freeze Thaw specimens suggest that the non-linear analysis of impact echo signals is capable of accurately quantifying distributed damage in concrete.