Impact of External Mechanical Loads on Coda Waves in Concrete

Abstract

During their life span, concrete structures interact with many kinds of external mechanical loads. Most of these loads are considered in advance and result in reversible deformations. Nevertheless, some of the loads cause irreversible, sometimes unnoticed changes below the macroscopic scale depending on the type and dimension of the impact. As the functionality of concrete structures is often relevant to safety and society, their condition must be known and, therefore, assessed on a regular basis. Out of the spectrum of non-destructive monitoring methods, Coda Wave Interferometry using embedded ultrasonic sensors is one particularly sensitive technique to evaluate changes to heterogeneous media. However, there are various influences on Coda waves in concrete, and the interpretation of their superimposed effect is ambiguous. In this study, we quantify the relations of uniaxial compression and uniaxial tension on Coda waves propagating in normal concrete. We found that both the signal correlation of ultrasonic signals as well as their velocity variation directly reflect the stress change in concrete structures in a laboratory environment. For the linear elastic range up to 30% of the strength, we calculated a velocity variation of −0.97‰/MPa for compression and 0.33%/MPa for tension using linear regression. In addition, these parameters revealed even weak irreversible changes after removal of the load. Furthermore, we show the time-dependent effects of shrinkage and creep on Coda waves by providing the development of the signal parameters over time during half a year together with creep recovery. Our observations showed that time-dependent material changes must be taken into account for any comparison of ultrasonic signals that are far apart in time. The study’s results demonstrate how Coda Wave Interferometry is capable of monitoring stress changes and detecting even small-size microstructural changes. By indicating the stated relations and their separation from further impacts, e.g., temperature and moisture, we anticipate our study to contribute to the qualification of Coda Wave Interferometry for its application as an early-warning system for concrete structures.

Identifying Damage In Concrete Using Coda Signals, Multi-Scale Simulations And Machine Learning

“…With further increase in loading, we see a reduction of dv v at the rate of 0.0041 MPa −1 due to diffuse microcracking. Overall, the trend reproduced from the simulation is similar to the experimental range reported in [1]. Moreover, a linear correlation between dv v and the overall precursor damage can be seen in Figure 7 once the precursor damage growth occurs.…”

“…It is essential to emphasize that the model's validity has been established through rigorous validation against experimental data. In our analysis, the microcrack parameters and material parameters of mortar matrix were adapted to match the material properties of concrete composition provided in [1] (concrete standard AB8 with cement type CEM I 42.5R with water/cement ratio of 0.45). In addition, a sub-model described in [6] is adapted to describe the stiffening effect related to microcrack closure observed in the elastic deformation regime [10].…”

“…the cell aspect ratio and cell growth rate (see [15]), these are calibrated parameters that control microcrack distribution in the cementitious matrix. In our simulation, the values were set to 1 18 , and 3.5 respectively.…”

Combinatorial Quantification of Multi-Features of Coda Waves in Temperature-Affected Concrete Beams

“…For the τ and ε of the coda wave, many factors such as stress and temperature will cause structural deformation, leading to the closure or development of micro-cracks, thereby changing the scattering wave path. These influences are reflected in the first two features [17,25,30]. However, after extracting the first two features, multivariate linear regression processing in high-dimensional space may be used to utilize different influencing factors having different spatial deflection directions to achieve decoupling between different influencing factors.…”

“…However, after extracting the first two features, multivariate linear regression processing in high-dimensional space may be used to utilize different influencing factors having different spatial deflection directions to achieve decoupling between different influencing factors. Study [30] also noted the need to separate different influencing factors. Compared to quantitative relationships established by a single parameter, the RMSE is reduced by 19% and 71% respectively, the SD is reduced by 25% and 73%, and the R 2 is increased from 0.93 and 0.88 to a level of 0.99 (beam B, the RMSE is increased by 24%, 86%; the SD is increased by 40%, 76%; and the R 2 is increased from 0.93 and 0.53 to a level of 0.99).…”