1. Introduction
Concrete is one of the most popular materials in civil engineering. Along with the increasing need for structural inspection, ultrasonic methods have been employed to evaluate the quality of materials in situ. Ultrasound is beneficial for characterizing the soundness of materials because the wave parameters such as velocity and amplitude are directly related to the mechanical properties of the materials [
1]. However, the contact-based methods have limited application due to the coupling procedure [
2]. First, the significant acoustic impedance mismatch between concrete and ultrasonic transducers made by piezoelectric materials requires a gel-type couplant, which is not practical in the field. Second, the surface roughness of concrete frequently causes irregular contact of the transducers. To avoid the multiple scattering effects by coarse aggregate, a low-frequency transducer below 100 kHz is typically used for standard concrete inspection applications [
3]. Moreover, the piezoelectric transducers usually have relatively large dimensions of the contact surface.
Recent advances in non-contact ultrasonic sensing technology have broadened the scope of nondestructive inspection of various engineering materials including concrete [
4]. Zhu et al., (2004) demonstrated that the leakage portion of surface waves measured by contactless sensors provides information on internal damage in concrete [
5]. Several studies have proved the effectiveness of the contactless sensing technology to inspect concrete structures [
6]. Neuenschwander et al., (2006) showed the measurement of leaky Rayleigh waves from mortar samples using a fully contactless ultrasonic system [
7]. Further developments in the contactless ultrasonic system field have been reported [
8,
9,
10], and unique signal processing algorithms have also been suggested in addition to the contactless ultrasonic hardware system [
11,
12,
13,
14]. As described in previous studies, the contactless ultrasonic system is an efficient method for the inspection of concrete structures, overcoming the drawbacks of the conventional contact-based approach.
On top of inspection applications (e.g., damage detection), the contactless ultrasonic system enables the provision of practical information on early-age material characteristics, such as the setting time. Conventional methods to determine a setting time are based on pin penetration tests, where a physical pressure load is applied to the surface of a concrete specimen. However, the methods recommend wet sieving to filter out coarse aggregate such as gravel, which is a significantly labor-intensive and time-consuming procedure. Choi et al., (2016) first demonstrated that the initiation of leaky Rayleigh waves is a useful tool to determine the final setting of early-age cementitious materials without physical contact with the specimens [
12]. Moreover, the incident angle of ultrasonic waves is a key parameter in defining the degree of the setting, where the larger angles provide the earlier Rayleigh wave initiation, resulting in a faster definition of the setting time. Hong et al., (2020) evaluated the effectiveness of contactless ultrasonic measurements in monitoring the stiffening behavior of concrete along with the maturity method [
15]. Hong and Choi (2021) further developed a contactless ultrasonic measurement hardware system and suggested a signal processing algorithm to investigate the elastic modulus of early-age cementitious materials [
16].
In the characterization of early-age concrete, the contactless ultra-sonic system showed great potential for field applications. Tran and Roesler (2020) applied the contactless ultrasonic system to the saw-cutting initiation time in concrete pavements and suggested the framework to predict the relationship between the saw-cut initiation time and final setting time [
17]. Saw cutting is required to control the quality of jointed plain concrete pavements to prevent early-age shrinkage cracks. Therefore, the contraction joints are constructed at the surface of the pavements using the saw cut. The saw cutting is a highly time-dependent procedure that causes raveling damage at a too early stage of concrete hardening or random cracking at a too late stage of concrete hardening. A proper saw-cut time window has been investigated based on the levels of produced damage using computer vision [
18,
19]. Tran and Roesler (2021) further evaluated the incident angle effect of the contactless ultrasonic system to identify the optimal setting definition for the saw cutting [
20]. Based on the previous studies, the delayed initiation time of leaky Rayleigh wave compared to the final setting is required to define the optimal saw-cutting time. Therefore, the smaller incident angle will give a better correlation between the optimal saw-cutting time and the initiation of leaky Rayleigh waves.
In this study, the optimal saw-cutting time is investigated using the concept of a contactless ultrasonic scheme. The uniqueness of this study over previous research lies in (1) the improvement of the sensor network of the contactless system using wireless data transfer for possible field applications, and (2) the development of a signal processing and data analysis scheme to determine the optimal saw-cutting time. Moreover, we optimized the critical testing parameters of the contactless ultrasonic system for application to saw-cutting time, such as the incident angle, measurement time intervals, and the number of sensor array elements. The theoretical background of leaky Rayleigh wave propagation in early-age concrete is supported using numerical simulations. Moreover, the developed hardware and signal processing algorithm were experimentally validated and compared to the conventional methods, such as the pin penetration and maturity methods.
2. Methodology
2.1. Theoretical Background of Leaky Rayleigh Wave Propagation in Early-Age Concrete
Rayleigh wave is one of the surface waves, propagating at the surface of a medium. The wave motion is elliptical rotation, which is caused by the surface guidance of body waves. The portions of longitudinal and shear waves generate the unique behavior of wave propagation. The relationship between material properties and Rayleigh wave velocity is given by
where
is the elastic modulus,
is the mass density, and
is Poisson’s ratio [
21].
At the joint-half space of air and concrete, small portions of the Rayleigh wave en0.ergy are leaked into the air (called leaky Rayleigh waves). A contactless ultrasonic measurement system measures the leaky Rayleigh wave without a physical coupling procedure to the surface of a concrete element. In non-destructive evaluation, the measurement of leaky Rayleigh waves has the great benefit of non-invasive material characterization.
Concrete is the composite material consisting of cement, and fine and coarse aggregate. The early-age concrete mixture is in a liquid state when water is added, and the hydration reaction makes the mixture harden in time. Conventionally, the degree of hardening at the initial stage is defined by the initial and final setting, generally representing the initiation of chemical adhesion and the physical stiffening of the mixture, respectively. The solid-state of concrete is considered as the time after the final setting. Leaky Rayleigh waves are measurable when the medium enables stress transfer, implying the propagation of both longitudinal and shear waves inside the medium. Therefore, the initiation of leaky Rayleigh waves represents the critical phase of shear resistance development in the medium, which is more relevant to the definition of the final setting.
The stiffening scenario with respect to leaky Rayleigh wave propagation is numerically demonstrated using the finite element method. The commercially available software, COMSOL Multiphysics, was used to analyze the propagation of the leaky Rayleigh wave at the joint-half space. The Multiphysics model included the air and concrete, where the ultrasonic excitation and sensing locations were positioned in the air. The details of the numerical simulation are described in
Table 1. Note that both materials were assumed to be homogeneous with respect to the applied wavelength, where the random scattering by aggregate was not considered. In addition, acoustic absorbing layers were placed around the model to eliminate reflection by the boundary, simulating infinite spaces. Therefore, the generated ultrasound was only guided by the joint between the air and concrete, implying the surface of the concrete. To investigate the stiffening phenomenon in concrete, a series of numerical simulations were carried out, where 26 cases of the elastic modulus of concrete ranging from 0.1 GPa to 25 GPa were considered in 1 GPa increments.
The simulation results presenting wavefield snapshots are shown in
Figure 1. The presented wavefield images were captured 300 microseconds after the ultrasonic excitation. The ultrasonic excitation was placed 50 mm apart from the surface of concrete acting as a point source shown on the left side of the images. At the beginning of hydration (
E = 0.1 GPa) shown in
Figure 1a, Rayleigh waves are not generated while others such as acoustic and body waves are identified in the air and concrete, respectively. After the modulus is developed to a certain amount (e.g.,
E = 1 GPa), Rayleigh waves propagate at the surface of concrete, and the corresponding leaky waves are observed in the air. In the hardened concrete case (
E = 25 GPa) seen in
Figure 1b, the body waves exhibit a faster propagation speed and a larger wavelength compared to those in fresh concrete seen in
Figure 1a. Also, the velocity of Rayleigh waves is much faster than that of the acoustics (
Figure 1b). Therefore, the contactless sensors receive the leaky Rayleigh waves first and direct acoustic waves later. Note that the amplitude of leaky Rayleigh waves is intrinsically lower than that of the acoustics. As the elastic modulus in concrete is developed, leaky Rayleigh waves propagate faster while the acoustic wave speed remains constant.
Based on the results of numerical simulations, the direct relationship between the leaky Rayleigh wave velocity and elastic modulus of concrete was derived as shown in
Figure 2. The velocity of leaky Rayleigh waves at each state of concrete was simply calculated from signals received by sensor arrays. The obtained relationship was presented as the dotted line in
Figure 2 while the theoretical range of Rayleigh waves based on Equation (1) is indicated in the shaded area. An equation (see
Figure 2) was derived from the trend line to estimate the elastic modulus directly from the measured leaky Rayleigh wave velocity. This simple equation is useful in practice, considering that information on Poisson’s ratio and mass density are typically lacking on construction sites. The suggested equation conservatively correlates the leaky Rayleigh wave velocity and the modulus of elasticity, which would be a good estimation of the elastic modulus of concrete in the field.
2.2. Sensor Network for Wireless Data Transfer and Signal Processing
To measure the leaky Rayleigh wave responses from concrete pavements, a wireless and contactless ultrasonic system was developed. The system includes two printed circuit boards (PCBs) encapsulated by an aluminum frame. Each PCB is designed for sensing and data acquisition parts, respectively. The sensing board consists of 16 MEMS (Microelectromechanical systems) ultrasonic sensor elements placed with 5 mm spacing. The data acquisition board is composed of several components, including a field-programmable gate array (FPGA), multiplexer, amplifier, and Wi-Fi module. The details of the wireless system are presented in
Figure 3.
The acoustic pressure is measured using the MEMS sensor array, and the collected analog signals are amplified 2000 times at the center frequency of 50 kHz. Then, the analog signals are converted to digital signals with a sampling rate of 2 MS/s and a dynamic resolution of 14 bits. The data acquisition process is repeated 16 times through each MEMS sensor element of the array, and the acquired data are wirelessly transferred to a personal computer. The stored data can be checked in real-time with programmed software, and measurement parameters such as the number of time-averaging can be set by an operator. The developed system is significantly improved in terms of the overall structure of the sensor network compared to the previously applied wired unit [
16]. First, wireless data transferring has great potential for field applications. The Internet of Things broadens the range of wireless data acquisition. The signal strength was evaluated as good (−50 to −60 dBm) with a maximum distance of 5 m [
22]. Second, the system is operated by a rechargeable battery, which is practical in pavement construction sites where electric power is frequently unavailable. Third, the number of sensors in the wireless system is twice that of the wired system, which increases the wavenumber resolution for data analysis in the frequency–wavenumber domain. With 16 MEMS sensors placed with 5 mm spacing, the length of the array is 80 mm, which is longer than the wavelength of Rayleigh waves (assuming 50 mm at 50 kHz) in hardened concrete.
In addition to the hardware development, we propose a signal processing approach to computing the leaky Rayleigh wave velocity from measured signals. An overview of the proposed signal processing approach is illustrated in
Figure 4.
First, bandpass filtering is applied to the measured signal set
to suppress noises other than the frequency band of excitation (40 to 50 kHz). Then, a two-dimensional (2-D) Fourier transform is applied to the bandpass filtered signal set
to convert the signal set from time–space (
t-x) to frequency–wavenumber (
f-k) domains, given by
where
is the converted
f-k domain signal set, and
j is the imaginary number. Next, a region of interest in the
f-k domain is highlighted by multiplying a filtering mask
and
, given by
where the filtering mask
is defined as
Note that the filtering operation shown in Equation (3) maintains wave components with the Rayleigh wave velocity of typical concrete while suppressing other unwanted components including noise. Then, peak frequency (
) and wavenumber (
) are obtained by picking
f and
k values at the maximum amplitude of
:
Finally, Leaky Rayleigh wave velocity (
) is computed as
Once
is obtained, the modulus of elasticity can be computed, given by
Note that Equation (7) was derived from the simulation results shown in
Figure 2. Monitoring the development of
or
E over time, the hardening process of early-age concrete can be monitored. Also, the optimal saw-cut timing can be determined using the
or
E data, which is presented in
Section 3.
Figure 4.
A flowchart of the proposed signal processing scheme.
Figure 4.
A flowchart of the proposed signal processing scheme.
2.3. Conventional Methods: Maturity and Pin Penetration
To characterize the state of early-age concrete, conventional methods such as maturity and pin penetration methods were also performed. The pin penetration method (ASTM C403) is a standard test method to determine the setting time of concrete mixtures [
23]. In this test method, pins having a different support area are used to measure the penetration resistance of the sieved mortar fraction from a concrete mixture, across different hardening stages of concrete. Dividing the applied load with the area of the used pin, a penetration resistance value is computed. Using the measured penetration resistance values, a smooth trend curve can be obtained, which visualizes the hardening of the tested concrete mixture. The initial and final setting times are determined by picking the time points that correspond to 3.5 MPa and 27.6 MPa, respectively.
Despite its wide acceptance as a means to determine the setting time of concrete, the pin penetration method has several limitations. First, performing the pin penetration test is highly labor-intensive. As the pin penetration method is not applicable directly to concrete with coarse aggregate, wet sieving needs to be carried out to extract the mortar fraction of the concrete mixture. Also, the test results can vary depending on the proficiency of the operator. To obtain reliable setting times, the operator needs to be trained for the test procedure. Moreover, the pin penetration test is typically carried out in a controlled laboratory environment that can be different from that of an actual construction site (e.g., temperature and relative humidity).
The maturity method (ASTM C1074) is a standard practice to estimate concrete strength using measured temperature data [
24]. Two maturity functions are widely used to compute the maturity index from the measured temperature history: (a) Nurse–Saul (NS) and (b) Freiesleben–Hasen and Pedersen (FHP) methods. The NS method is used to compute the temperature–time factor, and the FHP method to compute the equivalent age of concrete mixtures. In this study, the FHP method is used to compute the equivalent age given by
where
is the equivalent age at a specified temperature
,
Q the activation energy divided by the ideal gas constant, and
the average temperature of concrete during the time interval
. Typically, a calibration curve that relates between
and the compressive strength is constructed in a laboratory setting. Then, the calibration curve constructed for a given mixture design is used to predict the compressive strength of early-age concrete on site using the measured temperature data. In this study, we use the equivalent age
as an auxiliary index to help determine the optimal saw-cutting time, rather than to predict the compressive strength of concrete mixtures.
Like the pin penetration test, the maturity method has its own limitations. First, the laboratory curing temperature may not represent that of the concrete element on the actual construction site. Also, the strength of concrete cannot be estimated directly from the temperature measurement data; a calibration curve needs to be constructed before actual strength estimation is carried out, which is time-consuming. The maturity method does not capture local strength variations within a concrete element.