Ultrasonic Signal Characteristics in Pre- and Post-yield Steel Structures

The theory of acoustoelasticity has been the main concept behind most studies investigating the stress level in different structural materials. In this paper, an alternative approach for stress assessment is investigated whereby signal characteristics are identified under different stress levels. The results of an experimental investigation on longitudinal waves propagating perpendicular to the applied stress are presented. Ultrasonic signals were acquired from steel specimens subjected to different uniaxial tensile stress levels. Two well known Digital Signal Processing (DSP) methods; the Fast Fourier Transform (FFT) and Chirp-Z Transform (CZT), were used to analyse the signals in the frequency domain. Two other techniques; peak amplitudes and signal energies, were used for investigating the acquired signals in the time domain. This study confirms that the acoustoelastic effects on ultrasonic signals are miniscule within the elastic range for longitudinal signals travelling perpendicular to the applied stress. However, the results show a clear distinction between signal characteristics prior to and post yielding. Electronic Journal of Structural Engineering, 9 (2009) 2 in all of these studies for stress measurement. The sensitivity of this linear relationship was investigated for various stress and wave directions (Egle & Bray 1976). Figure 1 illustrates three possible wave directions for an element under tension. Figure 1. Possible orthogonal directions and designations of velocities in solids (Bray & Tang 2001). The first and the second indices of the velocities in Figure 1 represent the propagation of the wave and the direction of the movement of the particles (polarization direction) respectively. The velocities which have the same direction of wave propagation and polarization correspond to longitudinal waves (e.g. V11, V22), meanwhile others represent the velocities in a perpendicular plane, known as shear waves (e.g. V12, V13). The sensitivity of these waves to the material strain level is significantly affected by the polarization direction of the waves. As can be seen in Figure 2, the largest relative change in wave velocity is associated with longitudinal waves (V11) followed by shear waves when particles vibrate parallel to the applied load (V21). However, stress measurement with these two waves may not always be possible in practice where the placement of transducers for these velocities might not be practically possible. This is due to the fact that structures’ ends may be inaccessible, and if they are, distances between access points are usually long. Furthermore, stress levels usually vary along a structure’s length, which complicates the problem of identifying stress levels. Another important challenge is that the application of acoustoelasticity to measure the applied or residual stresses in terms of wave speed or time of flight (TOF) is not a reference-free technique since the distance between the source and the receiver of the ultrasonic wave needs to be known exactly (Junge et al. 2006). Furthermore, the dependency of ultrasonic velocity with stress can be significantly non-linear for some materials (Mishakin et al. 2006). Therefore, stress measurement using longitudinal and shear waves, where the particles vibrate perpendicular to the load requires more effective methods as the sensitivity of these waves to the stress state of the material is not significant. Figure 2. Sensitivity of longitudinal and shear waves to the strain (Egle & Bray 1976). Instead of using the relative changes in the wave velocity, this paper investigates the dependence of four different parameters of ultrasonic signals in time and frequency domains on the stress state of steel. Ultrasonic longitudinal signals, travelling perpendicular to the load (V22, based on the nomenclature in Figure 1), are acquired using a specially built testing system from steel specimens under uniaxial tension before and after yield under various stress levels. V22 was chosen in this study because it is the most likely velocity that can be acquired in the field from thin-walled steel structures because of accessibility issues. For the time domain analysis, the relationship of the applied stresses with the changes in the peak amplitudes and the signal energy values of the first three echoes of the acquired signals were investigated. The other two parameters are obtained using spectral analysis. The commonly used DSP techniques; namely the Fast Fourier Transform (FFT) and the Chirp-Z Transform, were used. All results were normalized with the values corresponding to the unstressed condition for eliminating the effect of different test settings. The dependence of the signal energy, time domain peak amplitudes, and FFT and CZT peak values on the stress level of the material are presented. The experimental results Electronic Journal of Structural Engineering, 9 (2009) 3 show that the effect of the stress state on the investigated signal characteristics may be used to detect yield in steel. 2 EXPERIMENTAL PROGRAM An experimental program was conducted to study ultrasonic signal characteristics before and after yield. The details of this program are described next. 2.1 Specimen preparation and properties All specimens were obtained from a quarter of an inch thick ASTM A36 steel plate. The plate was cut by a hydrocut waterjet machine to obtain sheet-type test specimens. The specimens have the dimensions of the rectangular sheet-type standard specimens following ASTM Standard E8-04 (ASTM 2004). Three initial tests were conducted to obtain information about the material’s mechanical properties in tension. For these tests, MTS 810 Hydraulic Materials Testing System was used, which was controlled by an MTS TestStar II controller that is programmed via a PC using MultiPurpose TestWare ® software (Model 793.10). The software allows the user to customize and generate special test procedures and store the test data from three channels, namely; force, displacement and strain. Two different MTS extensometers with one and two inch gage lengths were used for acquiring strain data. The results of these three material tests showed that the yield strength of the material is 310MPa (45ksi) while the ultimate strength is 469MPa (68ksi). Figure 3 shows the stress strain relationship for the specimen material. Figure 3. Stress-strain curve of steel used in this study.