Application Cases

Experiment Name: Corrosion Diagnosis Method for Rod-Shaped Components Based on Longitudinal Guided Waves

Research Direction: Nondestructive Testing

Testing Equipment: Signal generator, ATA-8202 power amplifier, data acquisition card, DC power supply, ultrasonic probe, steel rod, preamplifier.

Experimental Procedure:

Figure: Experimental Setup

The experimental setup is shown in Figure 3.2. The monitoring object was a steel rod under accelerated corrosion test conditions, with a total length of approximately 103 cm, of which the corroded section length was 94 cm. In the electrochemical accelerated corrosion test, a 3.5% sodium chloride solution was used as the electrolyte. An external DC power supply provided a constant current (350 mA) for the corrosion system. The steel rod was connected to the positive terminal of the power supply as the anode of the corrosion galvanic cell, and four equally spaced stainless steel plates were connected to the negative terminal of the power supply as the cathode of the corrosion galvanic cell. The NI data acquisition suite included a chassis, an arbitrary waveform generator, and an 8-channel digital signal acquisition board with a maximum sampling frequency of 60 MS/s. LabVIEW software was used to control the measurement system via a PC. The pitch-catch measurement method was adopted. Acoustic emission probes with a frequency response range of 0.1 MHz to 1.5 MHz were used as sensors to excite and receive pure longitudinal mode guided wave signals. The excitation signal was amplified using a radio frequency power amplifier (ATA-8202), and the received signal was amplified using a preamplifier. Designed fixtures were used to fix the probes at both ends of the steel rod under test, and oil was used as an interface couplant to enhance the transmission efficiency of vibration energy. During guided wave monitoring, the electrolyte was released through a tap on the plastic container, allowing the corroded section of the steel rod to be completely exposed to air. Meanwhile, the corrosion products accumulating on the steel rod surface over the course of the test were retained to simulate the actual corrosion state of the component.

The excitation signal was a 4-cycle Hanning window-modulated signal with a center frequency of 376 kHz. The relevant parameters of the wavelet basis function were set to F=5 and F=6 after debugging. The normalized received signal (for D-7mm) and its time-frequency analysis results are shown in Figure 3.3. The dispersion phenomenon of the received signal was very apparent and was in good agreement with the theoretical dispersion curve of the L(0,1) mode. Following the L(0,1) mode, there was a vibration persisting up to 780 μs at a frequency of 376 kHz, which was likely due to electromagnetic oscillations of the probe. Substituting the coordinates of point P, corresponding to the maximum wavelet coefficient (i.e., the maximum correlation between the wavelet basis function and the received signal), into the formula yielded an estimated steel rod diameter of 6.9943 mm, indicating that the estimated value was highly consistent with the theoretical result.

Figure: Guided Wave Received Signal

Figure 3.8 illustrates the guided wave received signals and their localized time-frequency analysis results for weight loss rates of 0%, 1.63%, 2.89%, and 4.13% during the early stage of the corrosion test.

Figure: Guided Wave Received Signals at Different Corrosion States

Experimental Results:
The interval estimation results of the remaining diameter of the steel rod as a function of corrosion progression, based on GUM (Guide to the Expression of Uncertainty in Measurement), are shown in Figure 3.9. When the corrosion weight loss rate was within 12%, the estimated diameter agreed well with the actual measured results. The uncertainty introduced by wavelet time-frequency analysis adequately covered and reasonably explained the deviation between the estimated and true values. However, as corrosion progressed, the remaining diameter tended to be progressively overestimated. Especially after the corrosion weight loss rate reached approximately 13%, the interval estimation results no longer covered the true value, and the remaining diameter assessment based on guided wave technology became unreliable. The assessment results could be considered "outliers." The appearance of outliers may be attributed to the introduction of new uncertainties due to changes in the measurement system during the later stages of corrosion, particularly the accumulation of corrosion products on the steel rod surface, which altered the guided wave propagation characteristics. The accumulating and densifying corrosion products increased the acoustic impedance, leading to greater leakage of guided wave energy into the corrosion product layer surrounding the steel rod matrix. This transformation changed the single-layer cylindrical waveguide into a double-layer cylindrical waveguide, consequently altering the dispersion characteristics of the guided waves. The portion of the wave propagating within the corrosion product layer exhibited a lower wave speed. Upon further superposition, the resulting guided wave speed in the double-layer cylindrical waveguide became lower than that in the corresponding single-layer cylindrical waveguide. Furthermore, the optimal wavelet correlation parameters in the time-frequency analysis were set based on the diameter measurement results of the steel rod in its pristine state. As the steel rod corrosion intensified, the received signal underwent significant distortion compared to the pristine state, which could potentially affect the results of the wavelet time-frequency analysis, subsequently introducing bias into the remaining diameter estimates.

The uncertainties inherent in the measurement process reduced the accuracy of corrosion assessment and could even lead to substantial deviations, thereby significantly underestimating the risks associated with corrosion evaluation. This is particularly critical for assessing uniform corrosion, where even a small weight loss due to corrosion could cause a substantial decline in the remaining performance of the component. When indirectly using the uniform corrosion amount to predict the remaining performance of the component, minor errors become amplified. In the following section, time series analysis methods will be further applied to process the guided wave monitoring results to obtain more accurate and reliable estimates of the remaining steel rod diameter.

Figure: Interval Estimation Results of Remaining Diameter Based on Guided Wave Monitoring Technology